peru@soleil.UUCP (Dave Peru) (11/15/88)
Definition of Intelligence: 1. Know how to solve problems. 2. Know which problems are unsolvable. 3. Know #1, #2, and #3 defines intelligence. This is the correct definition of intelligence. If anyone disagrees, please state so and why. If you take into account the unsolvable problems of Turing machines, then this proves Artificial Intelligence is impossible. "Artificial Intelligence" is an unsolvable problem. Human beings are not machines. Human beings are capable of knowing which problems are unsolvable, while machines are not.
joe@athena.mit.edu (Joseph C Wang) (11/15/88)
In article <484@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: >Definition of Intelligence: > >1. Know how to solve problems. >2. Know which problems are unsolvable. >3. Know #1, #2, and #3 defines intelligence. > >This is the correct definition of intelligence. If anyone disagrees, please >state so and why. Find an even number that is not the sum of two prime numbers. Find a set of integers a, b, c, and n in which a^n + b^n = c^n where n is greater than two. Are these problems solvable? Don't know? Must not be intelligent. -------------------------------- Joseph Wang (joe@athena.mit.edu) 450 Memorial Drive C-111 Cambridge, MA 02139
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (11/15/88)
In article <484@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes:
:Definition of Intelligence:
:
:1. Know how to solve problems.
:2. Know which problems are unsolvable.
:3. Know #1, #2, and #3 defines intelligence.
:
:This is the correct definition of intelligence. If anyone disagrees, please
:state so and why.
:
I disagree. Is a steelworker intelligent? Does a steelworker
know which problems are unsolvable (without being told)?
:
:Human beings are not machines.
:
Says who? Can you prove this? All the evidence I know points
toward human beings as being machines.
:Human beings are capable of knowing which problems are unsolvable, while
:machines are not.
What does knowing mean?
I thought Douglas Hofstedter's book put this argument to rest some
time ago. ok@quintus.uucp (Richard A. O'Keefe) (11/15/88)
In article <484@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: >Definition of Intelligence: > >1. Know how to solve problems. >2. Know which problems are unsolvable. >3. Know #1, #2, and #3 defines intelligence. > >This is the correct definition of intelligence. If anyone disagrees, please >state so and why. > (Gilbert Cockton is going to love me for this, I can tell...) Intelligence is a social construct, an ascription of value to certain characteristics and behaviours deemed to be mental. One child who has memorized the periodic table of the elements will be deemed intelligent, another child who has memorized baseball scores for the last N years will be deemed sports-mad, even though they may have acquired comparable bodies of information _by_comparable_means_. If we have three people in a room: Subject, Experimenter, and Informant, if Subject does something, and Informant says "that was intelligent", Experimenter is left wondering "is that a fact about Subject's behaviour, or about Informant's culture?" The answer, of course, is "yes it is". Dijkstra's favourite dictionary entry is "Intelligent, adj. ... able to perform the functions of a computer ..." (Dijkstra doesn't think much of AI...) In at least some respects, computers are already culturally defined as intelligent. >Human beings are not machines. I agree with this. >Human beings are capable of knowing which problems are unsolvable, while >machines are not. But I can't agree with this! There are infinitely many unsolvable problems, and determining whether a particular problem is unsolvable is itself unsolvable. This does _not_ mean that a machine cannot determine that a particular problem _is_ solvable, only that there cannot be a general procedure for classifying _all_ problems which is guaranteed to terminate in finite time. Human beings are also capable of giving up, and of making mistakes. Most of the unsolvable problems I know about I was _told_; machines can be told! Human beings are not machines, but they aren't transfinite gods either.
ap1i+@andrew.cmu.edu (Andrew C. Plotkin) (11/15/88)
/ Definition of Intelligence: / / 1. Know how to solve problems. / 2. Know which problems are unsolvable. / 3. Know #1, #2, and #3 defines intelligence. / / This is the correct definition of intelligence. If anyone disagrees, please / state so and why. Well, it seems to somehow miss the point... but I won't argue 'cause I don't know enough about intelligence. (Does anyone?) / If you take into account the unsolvable problems of Turing machines, then this / proves Artificial Intelligence is impossible. / / "Artificial Intelligence" is an unsolvable problem. / / Human beings are not machines. / / Human beings are capable of knowing which problems are unsolvable, while / machines are not. Wrong, wrong. I've had enough comp.sci. to have heard the proofs that certain problems are unsolvable. A notable fact: all those proofs were presented formally -- that is, the problem "is this (particular) problem Turing-solvable?" *was* Turing-solvable. There's also a misconception that humans really *can* solve all those Turing-insoluble problems. For example, the problem "Will this program always terminate for all inputs, or might it go into an infinite loop?" That's definitely Turing-insoluble, but can a human solve it? For *any* program given to him? I maintain that a human can be simulated by a Turing machine. Comments? --Z
arrom@aplcen.apl.jhu.edu (Ken Arromdee ) (11/15/88)
>Human beings are capable of knowing which problems are unsolvable, while >machines are not. Says who? -- "I don't care if you _did_ do it in a movie once, Gilligan is not breathing through that reed!" --Kenneth Arromdee (ins_akaa@jhunix.UUCP, arromdee@crabcake.cs.jhu.edu, g49i0188@jhuvm.BITNET) (not arrom@aplcen, which is my class account)
rolandi@gollum.UUCP (w.rolandi) (11/15/88)
In response to O'Keefe's: >>Human beings are not machines. >I agree with this. I too agree. Just the same, I would be interested in hearing your opinions as to how or why human beings differ from machines. Walter Rolandi rolandi@ncrcae.Columbia.NCR.COM NCR Advanced Systems Development, Columbia, SC
marty@homxc.UUCP (M.B.BRILLIANT) (11/15/88)
In article <484@soleil.UUCP>, peru@soleil.UUCP (Dave Peru) writes: > Definition of Intelligence: > > 1. Know how to solve problems. > 2. Know which problems are unsolvable. > 3. Know #1, #2, and #3 defines intelligence. Definitions are arbitrary. My only criteria for definitions are whether they create useful terms, and whether they conflict with previously accepted definitions. This one might be useful, conflicts with other definitions, but I'm willing to try it out. One problem with it is the use of the word ``know.'' There is no such think as ``knowing'' a definition abstractly. Definitions change, as people change the way they use words, because they change the way they use the ideas the words represent. A definition that is useful one day may be found later to be inadequate or self-contradictory. > This is the correct definition of intelligence. If anyone disagrees, please > state so and why. I disagree. There is no such thing as a correct definition, only useful or useless. The definition is interesting, at least, but, as noted above, it has problems. There is already a vague concept of ``intelligence'' waiting for a precise definition. The concept contains the notion that human behavior is characterized by some property called ``intelligence'' which is absent from the behavior of most other living things and most machines. That notion prejudices any attempt to define artificial intelligence, because it presupposes that machines are not intelligent. Any definition of ``artificial intelligence'' must allow intelligence to be characteristically human, but not exclusively so. > If you take into account the unsolvable problems of Turing machines, then this > proves Artificial Intelligence is impossible. > > "Artificial Intelligence" is an unsolvable problem. This is a statement of a proposition and a preface to a proof. It is not a proof. Proof required. > Human beings are not machines. Therefore..... what? Is that supposed to mean that all machines are bound by certain unsolvability rules derived from the study of Turing machines, but the human mind is exempt? Can that be proved? > Human beings are capable of knowing which problems are unsolvable, while > machines are not. Proof? It seems to me that all we humans have is an existence proof that some problems are unsolvable. Classifying all problems into solvable and unsolvable may be itself an unsolvable problem - does anybody know? We are, I think, back to the primitive notion that (a) intelligence is what only humans do, (b) machines are not human, hence (c) machines are not intelligent. M. B. Brilliant Marty AT&T-BL HO 3D-520 (201)-949-1858 Holmdel, NJ 07733 att!houdi!marty1 Disclaimer: Opinions stated herein are mine unless and until my employer explicitly claims them; then I lose all rights to them.
jsb@dasys1.UUCP (The Invisible Man) (11/16/88)
In article <7974@bloom-beacon.MIT.EDU> joe@athena.mit.edu (Joseph C Wang) writes:
:Find an even number that is not the sum of two prime numbers.
How about 2? Or do you consider 1 a prime?
:
:Find a set of integers a, b, c, and n in which a^n + b^n = c^n where n is
:greater than two.
How about a = b = c = 0?
:
:Are these problems solvable?
:
:Don't know? Must not be intelligent.
I won't say you're not intelligent; just careless.
--
Jim Baumbach uunet!actnyc!jsb
Disclaimer: The Tao that appears in this signature is not the eternal Tao.bradb@ai.toronto.edu (Brad Brown) (11/16/88)
Claiming that the existence of Turning-unsolvable problems precludes AI is just plain wrong. That old chestnut has been around for a long time and has been debunked many times, so here's my attempt to debunk it (again :-) Assumption 1: Turning machines are capable of computing any function that any imaginable computer (or what we know of as a computer) can compute. Therefore the limits of Turing machines apply to any computer that we can construct. Assumption 2: There are problems that Turing machines cannot compute. The canonical example is the halting problem -- there is a simple proof that it is impossible to write a program that will take as input another program and the input to the other program and decide whether the program will halt. There are other examples which can be proved to be non-computable because they can be related to (technically, "reduced to,") the halting problem. False assumption 3: Since Turing machines cannot compute everything imaginable, they can't compute (ie simulate) human intelligence. Assumption 3 is implicitly based on an assumption that artificial intelligence requires a solution to the halting problem. It's not at all clear that this is neccessarily the case. To show that a solution to the halting problem is required, you would have to show that intelligent beings could perform some task which was reducable to the halting problem. For instance, you would have to show me that you were able to determine whether a program would halt given the program and the input -- for any program, not just a specific one. Bet you can't think of *any* task humans perform that is *provably* halting-reducable. People have known for some time that there are limits to what can be computed by *any* mathematically-based system. The halting problem for Turing machines is only one example -- Godel's Incompleteness Theorem is a more general result stating that any axiomatic system of sufficient power will be fundimentally incomplete. (NB Systems that are not of 'sufficient power' are also not of sufficient interest to consider for the purpose of AI) Unfortunately, Godel's result is another example of a limitation of formal systems that is misunderstood to mean that formal systems cannot exhibit intelligence. In this case, Godel shows that there will exist theorems which are true within a system that the system will not be able to prove true. Again, the fallacy of the popular opinion is that intelligent systems will require that such truths be provably true. This is an even more tenuous claim than the requirement for solutions to the halting problem -- given the number of mistakes that humans make all the time, why should we expect that humans will have to know the truths of these pathalogical cases? Indeed, most of the "errors" that formal system will make in determining the truth of a theorem will occur only for pathological cases that are probably not important anyway. So, I regard claims that AI requires some form of unachievably "perfect" reasoning with great schepticism. I feel strongly that if you are a person who looks at the brain from the point of view of a scientist who examines causes and effects, then the possibility (though certainly not the achievability) of AI is very credible. IMHO, beliefs to the contrary imply a belief in a "magic" component to the brain that defies rational explanation. Sorry, but I just don't buy that. (-: Brad Brown :-) bradb@ai.toronto.edu +: O +: -+- Flame-retardant shielding --> +: | +: / \
nick@hp-sdd.hp.com (Nick Flor) (11/16/88)
In article <484@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: >Definition of Intelligence: > >1. Know how to solve problems. Okay, my calculator (an HP28S) can do this. >2. Know which problems are unsolvable. My 28S can do this too. (It beeps when I ask it to solve an unsolvable equation). >3. Know #1, #2, and #3 defines intelligence. The only thing interesting about this last statement is that it's recursive. Well... my 28S can even do recursion. So, by your definition, my calculator is intelligient. But by current standards of intelligience, it isn't. Therefore your definition is wrong. >"Artificial Intelligence" is an unsolvable problem. But is the creation of intelligience an unsolvable problem? Nah, people make babies all the time. Nice try. Nick -- + Disclaimer: The above opinions are my own, not necessarily my employer's. + + Oh, sure. Sure. I'm going. But I got | Nick V. Flor * * o * * + + your number, see? And one of these | Hewlett Packard SDD * * /I\ * * + + days, the joke's gonna be on you. | ..hplabs!hp-sdd!nick * * / \ * * +
peru@soleil.UUCP (Dave Peru) (11/16/88)
In article <484@soleil.UUCP> I write: >>Definition of Intelligence: >> >>1. Know how to solve problems. >>2. Know which problems are unsolvable. >>3. Know #1, #2, and #3 defines intelligence. There is a misunderstanding what I meant by this statement, especially #2. Human beings KNOW the "halting problem for Turing machines", my point is that machines can NOT know the "halting problem for Turing machines". Please describe how you would give this knowledge to a computer. All uncomputability problems come from dealing with infinity. Like naive set theory, naive Artificial Intelligence does not deal with paradoxes and the concept of "infinity". Human beings understand the concept of "infinity", most of mathematics would be meaningless if you took out the concept of "infinity". Mathematicians would be quite upset if you told them that they were really fooling themselves all this time. Physicists use "infinity" for predicting reality. Finite machines cannot understand "infinity". For the concept of "infinity" to have any meaning at all you MUST have the computational strength of reality.
ap1i+@andrew.cmu.edu (Andrew C. Plotkin) (11/16/88)
/ Human beings KNOW the "halting problem for Turing machines", my point is / that machines can NOT know the "halting problem for Turing machines". / Please describe how you would give this knowledge to a computer. / All uncomputability problems come from dealing with infinity. / Like naive set theory, naive Artificial Intelligence does not deal with / paradoxes and the concept of "infinity". / Finite machines cannot understand "infinity". Finite machines can understand both infinity and the halting problem the same way they can understand any other theorem in a formal system. Give it the definitions and axioms, and let it crank out theorems. "The halting problem is Turing-insoluble" is a theorem that's not hard to prove. Program to do this sort of thing have been written. Hell, that's how humans did it. "Infinity" is a single concept; you don't need an infinite amount of information to comprehend it or use it. You only need the axioms and definitions that mention it. / For the concept of "infinity" to have any meaning at all you MUST have the / computational strength of reality. In mathematics, the concept of "infinity" has no meaning at all. In "real life," it is useful in describing certain aspects of the universe, but no more useful than the concept "two". A program which can handle one can handle the other. --Z
bjornl@blake.acs.washington.edu (Bjorn Levidown) (11/16/88)
In article <490@soleil.UUCP> it is stated that >For the concept of "infinity" to have any meaning at all you MUST have the >computational strength of reality. What is the computational strength of reality and what is inherent in a computer system which forces this limitation? Why can the human brain/mind comprehend reality? It is, after all, only a biological computer of a sort. Bjorn Levidow bjornl@blake.UUCP University of Washington Department of Psychology, NI-25
ap1i+@andrew.cmu.edu (Andrew C. Plotkin) (11/16/88)
/:: Find a set of integers a, b, c, and n in which a^n + b^n = c^n where n is /:: greater than two. /:: /:: Are these problems solvable? /:: /:: Don't know? Must not be intelligent. / / How about a = b = c = 0? / I won't say you're not intelligent; just careless. Joking aside, I'd sort of expect an AI to be as careless as a human... --Z
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (11/16/88)
In article <1738@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >Says who? Can you prove this? All the evidence I know points >toward human beings as being machines. Can't know much then. See C18 enlightenment debate, Descartes, goals of encylopaedists, limitations of enlightenment rationality (unfortunately ossified in American political values). This is at least a 300 year old debate (not counting earlier millenium debate on free will, and still going strong). I cannot see how an educated (and intelligent :-)) person could possibly be so ignorant of the cultural context of mechanistic models of humans. Looking at other US intellectual traditions like functionalism and sociobiology, it's no surprise though. All machines are artefacts. Humans are not artefacts. Humans are not machines. Humanity and culture are inseparable. Techies are uncultured. Techies are inhuman. Don't flame me, it was only my deterministic logic which infered this :-) :-) :-) :-) :-) :-) :-) :-) :-) :-) :-) :-) :-) :-) :-) :-) -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert
ok@quintus.uucp (Richard A. O'Keefe) (11/16/88)
In article <490@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: >In article <484@soleil.UUCP> I write: >>>Definition of Intelligence: >>>1. Know how to solve problems. >>>2. Know which problems are unsolvable. >>>3. Know #1, #2, and #3 defines intelligence. >There is a misunderstanding what I meant by this statement, especially #2. >Human beings KNOW the "halting problem for Turing machines", my point is >that machines can NOT know the "halting problem for Turing machines". >Please describe how you would give this knowledge to a computer. I was afraid for a minute there that you were going to say "know how to solve _practical_ problems, have a _practical_ grasp of which problems are _feasible_", but no, it's the halting-problem type of problem after all. What does it mean to say 'Human beings KNOW the Halting Problem'? As a plain matter of fact, most of them _don't_. I think I do, but what I mean by that is that I have enough formal knowledge about mathematical logic to follow the books, to relate some of the key concepts to each other, and to deploy this information in further formal reasoning. I do _not_ know, for any given program, whether or not it halts until I have examined that particular program, and even then the Law I _really_ rely on is Murphy's 1st law of bugs: "There is a mistake in this program." Boyer & Moore have a paper on a machine-generated proof of Goedel's Theorem. Read that paper to see how to "give this knowledge to a computer". Getting a computer to "know" mathematical things is a Small Matter of Programming. >All uncomputability problems come from dealing with infinity. In a strict sense, yes. But you can find oodles of NP-hard problems without thinking once about infinities, and NP-hard is just as good as uncomputable for most practical purposes. >Like naive set theory, naive Artificial Intelligence does not deal with >paradoxes and the concept of "infinity". There have been several conferences and workshops on knowledge representation. There is no shortage of papers discussing paradoxes in these areas. Smullyan has a fine paper on some paradoxes of belief. Not all AI work is naive. (There is an infinite regress in "I think that he thinks that I think ..." which has to be headed off; this _has_ occurred to people.) Surely set theory has taught us by now that there is no such thing as THE concept of infinity: omega is not Aleph-null is not the-point-at- infinity is not ... is not the common-sense notion of infinity. >Human beings understand the concept of "infinity", most of mathematics would >be meaningless if you took out the concept of "infinity". As for the second clause, clearly you are not an Intuitionist. As for the first, this is simply false: the vast majority of human beings have not had the technical training to distinguish between omega, Aleph-null, and the reciprocal of an infinitesimal in non-standard arithmetic, and those who _have_ had such training would probably be humble enough to admit that we are still nibbling at the edges of the concepts. >Finite machines cannot understand "infinity". Why not? The whole human race has so far encountered only a finite number of facts about the various infinities. By starting this joust in a FORMAL field you've lost the game in advance. >For the concept of "infinity" to have any meaning at all you MUST have the >computational strength of reality. For the concept of infinity to have any meaning at all to WHOM? If you mean "in order to understand infinity correctly, the understander must have the computational strength of infinity", maybe, but it is not clear to me that any human being understands infinity that well, particulary not one who talks about THE concept of infinity. What _is_ "the computational strength of reality", and how is it possible for finite beings to possess it? [Someone who believes that we are god could consistently believe that humans are not finite bounded creatures and so can "know" infinity. New AIge? On the other hand, a god so easily deceived could be wrong...]
kers@otter.hpl.hp.com (Christopher Dollin) (11/16/88)
Dave Peru says: | Human beings KNOW the "halting problem for Turing machines", my point is | that machines can NOT know the "halting problem for Turing machines". | | Please describe how you would give this knowledge to a computer. The same way I would give the knowledge to a student. Oh, machines aren't intelligent? But isn't that what we're discussing already? | All uncomputability problems come from dealing with infinity. Hm. Wouldn't "unboundedness" be a better term to use than "infinity"? [Since you have elsewhere claimed said "zero = infinity" isn't against your intuition, doesn't it follow that computability problems come from dealing with zero too?] | Like naive set theory, naive Artificial Intelligence does not deal with | paradoxes and the concept of "infinity". But does that matter? | Human beings understand the concept of "infinity", most of mathematics would ^^^^^^^^^^^^ | be meaningless if you took out the concept of "infinity". Mathematicians | would be quite upset if you told them that they were really fooling themselves | all this time. Physicists use "infinity" for predicting reality. *Some* human beings do. Some don't. Some mathematicians have a downer on "infinity". | Finite machines cannot understand "infinity". Hm. I claim human beings are counterexamples, but you don't believe that humans are machines. [Probably none of us agree on what "machine" means anyway. We might have better agreement on "human".] Do you claim that an infinite concept can't be captured in a finite way? I would have thought that many of our early notions of infinity were derived by a uniform extension of finite patterns ... to borrow the book title, "One ... Two ... Three ... Infinity". | For the concept of "infinity" to have any meaning at all you MUST have the | computational strength of reality. Is reality finite? ["Is there an odd or even number of elementary particles in the universe?" - no I *don't* propose to reopen that flamefeast.] What is "computational strength"? Are machines unreal? Is the filling in my sandwiches determined only when I bite through the bread? When will the long dark teatime of the soul be a paper back? Regards, | "Once I would have been glad to have made your acquaintance Kers. | Now I feel the danger brought about by circumstance."
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (11/16/88)
In article <490@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: > >Finite machines cannot understand "infinity". > Are human beings not finite? Are they not machines? If they are not machines, then how do they differ? What is it about this difference that allows us to understand infinity? Since you made this assertion, I must assume you have some belief or evidence to back it up, would you please provide it? If the answer is "I don't know", then how do you know that the essence of understanding infinity isn't mechanical? >For the concept of "infinity" to have any meaning at all you MUST have the >computational strength of reality. I don't understand this statement. What computational strength does reality have? Are computers not as real as humans?
anderson@secd.cs.umd.edu (Gary Anderson) (11/17/88)
In article <490@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: >In article <484@soleil.UUCP> I write: > >All uncomputability problems come from dealing with infinity. > >Like naive set theory, naive Artificial Intelligence does not deal with >paradoxes and the concept of "infinity". > >Human beings understand the concept of "infinity", most of mathematics would >be meaningless if you took out the concept of "infinity". Mathematicians >would be quite upset if you told them that they were really fooling themselves >all this time. Physicists use "infinity" for predicting reality. There are very many different concepts of infinity even within mathematics. I do not know the relevant comp sci literature, but I would be surprised if you could not in some way model or otherwise provide some of these concepts to a program designed to prove theorems in certain contexts. It seems to me that the power of the concept of infinity lies in how its is used to solve or to characterize the solution of certain questions in mathematics , and that so far as mathematics is concerned it does not have an existential basis independent of how it is or ought to be used. Anecdote: Professor Garrett Birkhoff, a prominent mathematician not unfamiliar with the use of infinity in mathematics, often remarks in his undergraduate ODE class that when mathematicians begin to speak about infinity, they don't know what they are talking about. From this observation, I suggest another difference between man and machine. Perhaps man is just better at coping with the "reality" of not "knowing".
jbn@glacier.STANFORD.EDU (John B. Nagle) (11/17/88)
We already had this discussion this year. We had it at least twice last year. So can it. John Nagle
hassell@tramp.Colorado.EDU (Christopher Hassell) (11/17/88)
Upon jumping into this fray, first I might notice that so much of this
(quite logically so) is the discussion of the meaning of intelligence as
we try to get it *NOT* related to humans. (We are our own first model it seems.)
I suggest that we cannot approach one that is not embedded with human implica-
tions so it might be a better idea to *somehow* find another word, being that
intelligence is too loaded to formalize.
Here goes the main stuff:
The most ticklish of the issues that confront formalizing intelligence is simply
an observation that we (*humans*) seem innately different from computers
(*machines*) and we, as with any good guess, seek to support it, being it has
some apparently solid stuff to support it.
I think this issue boils down to something else : -Determinism-. The typical
picture and one that I, myself, note causes revulsion is that of us as machines
which seems to imply deterministic and with predictable boundaries (finite),
and ones without ability to modify. (This is usually the issue in alot of sci-
fi on the subject)
Determinism in humans is always a fun issue, but also a irritatingly difficult
one to REALLY nail down. In other words, we ain't. At the nth degree, we
aren't predictable so no one need worry about the limits on their thoughts.
Other issues like seeing the future or the general flow of things [fate!!]
always seem relative to the person believing them and are therefore moot.
The hypothesis that seems to make sense is that, given those noisy neurons
and their interconnections, CHAOS seems to be the logical assumption at
some level (hopefully a lower one, well for some people |-)
On to more meaningful stuff, other issues are that if we are machines where
can we arrive at conclusions about ourselves and also where do we this knowledge
of the world? That would seem to come from the fact that any answer lies in a
*complete* definition of the problem. So given problems we can find answers,
because the problems have the arbitrary information. (all of the above
supposes a closure theorum on information as a substance, it can only be lost
supposedly (the world is full of chaos 'making' more?))
Given this it seems natural that if we get at *SYMBOLIC* scheme of
observation we can adapt to learning anything within our grasp,
(worked out pretty neat huh?) and that includes introspection
(wierd, but still symbolizible).
------THE LAST POINT-------
The MOST ticklish concept left would now be old hat.
We started with stuff like believing we didn't have the only nice
flat perspective (i.e. heavenly bodies existing, w/round earth)
The next general milestone was our conceding that we *ain't* at the
center of our universe either (This one's still going)
After this was a shocker, we are not separate from animals and the
`unintelligent' entities we see around us. (Tough one)
The last may be that we can derive the instructions for making an inherently
large part of what we call in ourselves `human', (`thinking beings')
What this may lead to if the link is proven, is the most fun of all. (ugh)
if our self-proclaimed `essance' is creatible and finite and we now know the
tools are left out that could make another for ourselves ... egad
What, completely defined, would we create once we figure out how to use them????
(being the notibly imperfect and selfish creatures we are!!)
If you've persevered this far through the text, you must either have a flame in
mind or have fallen asleep thumping the space bar |^/ zzzzzz
-------------------------------------------------------------------------------
YES, more mental masturbation for use in your home or office!!
provided by (without responsibility of course! He was probably possessed at the
time. What else could explain it?) >>>>C>>>H>>>>>
...ncar!boulder!tramp!hassell (and oh so much of it)
-------------------------------------------------------------------------------fransvo@htsa (Frans van Otten) (11/18/88)
In article <88Nov15.170837est.707@neat.ai.toronto.edu> bradb@ai.toronto.edu (Brad Brown) writes: >of AI is very credible. IMHO, beliefs to the contrary imply ^^^^ What does IMHO stand for ? -- Frans van Otten Algemene Hogeschool Amsterdam Technische en Maritieme Faculteit fransvo@htsa.uucp
fransvo@htsa (Frans van Otten) (11/18/88)
In article <484@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: >Human beings are not machines. I don't know how important intuition etc. is for (human) intelligence, nor do I know much about neurology, but viewing the human brain as a lot of 'biological transistors' (neurons etc.), in what way do we differ from a computer ? Why would it be impossible to make a machine, maybe as big as the Wolrd Trade Center, that performs (part of) the functions the human brain (small as it is) performs ? Besides, there is not just one kind of intelligence. -- Frans van Otten Algemene Hogeschool Amsterdam Technische en Maritieme Faculteit fransvo@htsa.uucp
dhw@itivax.UUCP (David H. West) (11/18/88)
In article <88Nov15.170837est.707@neat.ai.toronto.edu> bradb@ai.toronto.edu (Brad Brown) writes: >Godel's Incompleteness Theorem is a more general result >stating that any axiomatic system of sufficient power will >be fundimentally incomplete. (NB Systems that are not of >'sufficient power' are also not of sufficient interest to >consider for the purpose of AI) That's by no means obvious. Formal reasoning is a relatively recent addition to the human behavioral repertoire; people do it neither naturally (it takes them until well into adolescence to become halfway competent, whereas they walk and talk adequately ten years earlier) nor well (look at all the incorrect 'proofs' that get published), and most people manage perfectly well most of the time without using it at all (see the work of Kahneman and Tversky). Let's not get too logicocentric when talking about intelligence. It's at least plausible that nature wouldn't produce complete implementations of computationally problematic paradigms if there were a cheaper but adequate alternative. (Cf. Simon's satisficing, optical illusions etc.) >[...] Godel >shows that there will exist theorems which are true within >a system that the system will not be able to prove true. [...] >a belief in a "magic" component to the brain that defies >rational explanation. Sorry, but I just don't buy that. Eh? You have to admit the possibility, because you've just declared your belief in theorems which are true-but-unprovable in a system. If that isn't "defying rational explanation" (within the system), I don't know what is. -David West dhw%iti@umix.cc.umich.edu {uunet,rutgers,ames}!umix!itivax!dhw CDSL, Industrial Technology Institute, PO Box 1485, Ann Arbor, MI 48106
ok@quintus.uucp (Richard A. O'Keefe) (11/18/88)
In article <4714@boulder.Colorado.EDU> hassell@tramp.Colorado.EDU (Christopher Hassell) writes: >The MOST ticklish concept left would now be old hat. > We started with stuff like believing we didn't have the only nice > flat perspective (i.e. heavenly bodies existing, w/round earth) That has indeed been old hat for >2200 years. > The next general milestone was our conceding that we *ain't* at the > center of our universe either (This one's still going) Note that Ptolemaic astronomy did not put the Earth at the centre of the universe & that the mediaeval perspective could more accurately be described as "at the bottom of a pit" than "at the centre of the universe". > After this was a shocker, we are not separate from animals and the > `unintelligent' entities we see around us. (Tough one) There is a fascinating book called "The Criminal Prosecution and Execution of Animals" (that's from memory, the book is at home) which was republished a couple of years ago. Up until about the 18th century, it turns out that it was a common occurrence for animals (even a swarm of locusts) to be tried in court. This doesn't sound as though premodern Europe thought there was such a great separation. How can you possibly try a sow which ate its piglets for murder unless you believe that it was a responsible moral being? History is not only stranger than we imagine, it is _other_ than we imagine.
smryan@garth.UUCP (Steven Ryan) (11/18/88)
>Professor Garrett Birkhoff, a prominent mathematician not unfamiliar with the >use of infinity in mathematics, often remarks in his undergraduate ODE class >that when mathematicians begin to speak about infinity, they don't know >what they are talking about. Just a short note that is a mortal sin for a mathematician to say `inf***ty.' Penance usually is usually five Hail Davids. The word `infinite' only occurs before `set,' `infinite set,' where it has a very precise meaning. The correct term is `arbitrary.' Discrete transistion machines with arbitrary resources cannot necessarily navigate through a partial recursive set in finite time. Whether humans can is an open question. -- -- s m ryan +---------------------------------------+--------------------------------------+ | ....such cultural highlights as | Nature is Time's way of having | | Alan Thicke, and, uh,.... | pausible deniability. | +---------------------------------------+--------------------------------------+
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (11/18/88)
In article <0XTukNy00Xol41W1Ui@andrew.cmu.edu> ap1i+@andrew.cmu.edu (Andrew C. Plotkin) writes: >I maintain that a human can be simulated by a Turing machine. Comments? OK, propose a Turing Machine configuration that will emulate the decision making which made Quayle candidate for vice-president :-) (Mickey Mouse is rumoured to wear a Dan Quayle wrist-watch, BBC radio 4 morning news!) -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (11/18/88)
In article <1654@hp-sdd.HP.COM> nick@hp-sdd.hp.com.UUCP (Nick Flor) writes: >But is the creation of intelligence an unsolvable problem? >Nah, people make babies all the time. Babies are pretty dumb. Left to themselves, their MTBF is about a day, and then you don't get another chance to measure it. Intelligence arises through socialisation. No-one can guarantee to give any child a given level of 'intelligence' by the age of 18, although the bases of educational attainment are fairly well understood. Leave a child to themself and they will develop little of what non-AI types call intelligence (although they will pick up the blue block). This is a fact, as a few very miserable individuals have had the misfortune to prove. -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert
bradb@ai.toronto.edu (Brad Brown) (11/18/88)
In article <392@itivax.UUCP> dhw@itivax.UUCP (David H. West) writes: >In article <88Nov15.170837est.707@neat.ai.toronto.edu> bradb@ai.toronto.edu (Brad Brown) writes: >>Godel's Incompleteness Theorem is a more general result >>stating that any axiomatic system of sufficient power will >>be fundimentally incomplete. (NB Systems that are not of >>'sufficient power' are also not of sufficient interest to >>consider for the purpose of AI) > >That's by no means obvious. Formal reasoning is a relatively recent >addition to the human behavioral repertoire... I conceed that I don't have the backround to support this statement, but my article was a followup to a proposal that because Turing machines could not compute some functions (eg the halting problem) no computer could be made to compute "intelligence." I was arguing that the incompleteness of formal systems is not a sufficient to proof that formal systems cannot be intelligent. >>[...] Godel >>shows that there will exist theorems which are true within >>a system that the system will not be able to prove true. >[...] > >>a belief in a "magic" component to the brain that defies >>rational explanation. Sorry, but I just don't buy that. > >Eh? You have to admit the possibility, because you've just declared >your belief in theorems which are true-but-unprovable in a system. >If that isn't "defying rational explanation" (within the system), I >don't know what is. NO!! All Godel's theorem says is that every sufficiently powerful formal system will not be able to prove the truth of some true theorems. I implied earlier in my article that I felt the brain could not prove these truths *either*, so there is no contradiction. (-: Brad Brown :-) bradb@ai.toronto.edu
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (11/18/88)
From article <392@itivax.UUCP>, by dhw@itivax.UUCP (David H. West):
" In article <88Nov15.170837est.707@neat.ai.toronto.edu> bradb@ai.toronto.edu (Brad Brown) writes:
" >Godel's Incompleteness Theorem is a more general result
" >stating that any axiomatic system of sufficient power will
" >be fundimentally incomplete. (NB Systems that are not of
" >'sufficient power' are also not of sufficient interest to
" >consider for the purpose of AI)
"
" That's by no means obvious. Formal reasoning is a relatively recent
" addition to the human behavioral repertoire; people do it neither
" naturally (it takes them until well into adolescence to become
And for that matter, we can do a certain amount of formal reasoning
within the bounds of a complete system, since predicate logic
is complete. Goedel published a proof in 1930.
Greg, lee@uhccux.uhcc.hawaii.edumarty@homxc.UUCP (M.B.BRILLIANT) (11/19/88)
In article <17847@glacier.STANFORD.EDU>, jbn@glacier.STANFORD.EDU (John B. Nagle) writes: > > We already had this discussion this year. We had it at least > twice last year. So can it. Interesting comment. The title of the newsgroup is an abbreviation for ``artificial intelligence.'' Its subject is ``artificial intelligence.'' The discussion is about what it means when you juxtapose the two words ``artificial'' and ``intelligence.'' The poor sap who started it wanted to know whether he had the right definition of intelligence. Did he? If we had the same discussion already this year, and twice last year, did we settle it? What was the outcome? Do we know what ``artificial intelligence'' is? Or are we just talking about something we don't know anything about? I am ready to allow that ``artificial intelligence'' could be a word all by itself, which does not derive its meaning from the two separate words ``artificial'' and ``intelligence.'' But I certainly don't want to reopen a discussion that has already been settled. Just somebody tell us what the answer is. M. B. Brilliant Marty AT&T-BL HO 3D-520 (201) 949-1858 Holmdel, NJ 07733 att!houdi!marty1 Disclaimer: Opinions stated herein are mine unless and until my employer explicitly claims them; then I lose all rights to them.
mark@verdix.com (Mark Lundquist) (11/19/88)
In article <484@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: >Definition of Intelligence: > >1. Know how to solve problems. >2. Know which problems are unsolvable. >3. Know #1, #2, and #3 defines intelligence. > >This is the correct definition of intelligence. If anyone disagrees, please >state so and why. OK, I'll bite. First of all, in regard to (2), since there is no algorithm for deciding whether or not an arbitrary problem is solvable, it's hard to see how anything could posses property (2). Also, I don't understand why you include (3). What claims does it allow you to make about intelligent agents or intelligent behavior? What would you say about a being that possessed (1) and (2) but not (3)? Each of (1), (2), and (3) is formulated in terms of 'knowing'. Just what do you mean by "knows"? It seems possible that there is a sense of 'knowing' such that if a being could be truly said to 'know' even one thing in that sense, that being would be intelligent. As for proclaiming the above definition to be "correct", again it's not clear to me what you mean. You need to show that your definition is what people mean (or at least ought to mean) when they use the word "intelligent". Your definition certainly doesn't seem to be correct in this sense. You might choose to propose this definition as that of some particular _species_ of intelligence (I don't know why you would; I cofess that I'm quite unable to imagine what such a species of intelligence would be like), but I don't think in that case that you would have arrived at any useable concept of intelligence. This leads to my final point, which is that there's a significant burden-of-proof issue. Defining things is tricky business. There are predicates that we can and do use meaningfully but which don't appear to have any correct definition. Such terms are normative rather than descriptive, and they appear to be most adequately formulated in terms of "what it isn't" than "what it is" (I personally suspect that "intelligent" is one of these terms). When proposing a definition of something, especially a definition as problematic as this one, a little explaining of how you arrived at the definition is in order. You might try to anticipate some of the difficulties with it, and show why you believe that your definition isn't specious. As it is, the definition is little better than (4) Prefers pink grapefruit to yellow grapefruit (5) Stares mindlessly at toothpicks (6) Knows that (4), (5), and (6) constitute intelligence "This is right and if you don't think so, say why".
ap1i+@andrew.cmu.edu (Andrew C. Plotkin) (11/19/88)
/ In article <1738@cadre.dsl.PITTSBURGH.EDU>
/ geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes:
/ >Says who? Can you prove this? All the evidence I know points
/ >toward human beings as being machines.
/
/ Can't know much then.
/ See C18 enlightenment debate, Descartes, goals of encylopaedists,
/ limitations of enlightenment rationality
I know some Descartes, little of the others. However, I know a lot of human
beings, and I've never seen evidence that they (or I) are not machines.
/ I cannot see how an educated (and intelligent :-)) person could
/ possibly be so ignorant of the cultural context of mechanistic models of
/ humans.
What does cultural context have to do with it? But if they've shown evidence
that human beings aren't machines, by all means post a summary -- many many AI
researchers will be fascinated.
/ All machines are artifacts.
/ Humans are not artifacts.
/ Humans are not machines.
Fallacy there. You should know better.
--Znumork@ndsuvax.UUCP (James Mork) (11/20/88)
> >Dijkstra's favourite dictionary entry is > "Intelligent, adj. ... able to perform the functions of a computer ..." >(Dijkstra doesn't think much of AI...) > Does anyone besides myself have this almost cynical hatred for Dijkstra and the way that the Computer Science community revers him? I have seen so many algorithms atributed to Dijkstra that have been common solutions in everyday problems for thousands of years. It makes me want to throw up. -- UUCP Bitnet Internet uunet!ndsuvax!numork numork@ndsuvax numork@plains.nodak.edu
ok@quintus.uucp (Richard A. O'Keefe) (11/20/88)
In article <1808@ndsuvax.UUCP> numork@ndsuvax.UUCP (James Mork) writes: > Does anyone besides myself have this almost cynical hatred for > Dijkstra and the way that the Computer Science community reveres > him? I have seen so many algorithms attributed to Dijkstra that > have been common solutions in everyday problems for thousands > of years. It makes me want to throw up. From Funk&Wagnall's dictionary: cynical: Distrusting or contemptuous of virtue in others; sneering; sarcastic. [The] doctrine [of the Cynics] came to represent insolent self-righteousness. If you have seen any "algorithms attributed to Dijkstra" which "have been common solutions in everyday problems for thousands of years", NAME THEM. Ok, so Dijkstra is a Dutch curmudgeon with a low tolerance for hype, and has in consequence little liking for AI (but not robotics). That's a reason for throwing up? Not on _my_ terminal, thank you!
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (11/20/88)
From article <1919@garth.UUCP>, by smryan@garth.UUCP (Steven Ryan): " >Professor Garrett Birkhoff, a prominent mathematician not unfamiliar with the " >use of arbitrary in mathematics, often remarks in his undergraduate ODE class " >that when mathematicians begin to speak about arbitrary, they don't know " >what they are talking about. " " Just a short note that is a mortal sin for a mathematician to say " `arb***ry.' Penance usually is usually five Hail Davids. " " The word `infinite' only occurs before `set,' `infinite set,' where it has " a very precise meaning. " " The correct term is `arbitrary.' Now it makes even more sense.
morrison@grads.cs.ubc.ca (Rick Morrison) (11/21/88)
In article <4264@homxc.UUCP> marty@homxc.UUCP (M.B.BRILLIANT) writes: > ... Do we know what ``artificial intelligence'' is? Or are we just talking > about something we don't know anything about? ... Just somebody >tell us what the answer is. The answer is "who cares?" Does anyone in this group actually _do_ AI? I'm beginning to think that the most appropriate definition of AI is "a discipline concerned with the uninformed examination of unresolvable philosophical and psychological issues."
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (11/21/88)
In article <1908@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: > >Intelligence arises through socialisation. > Why is this a good argument against the possibility of machine intelligence? If a large enough neural network could be created, could it also not receive socialization a la HAL?
doug@feedme.UUCP (Doug Salot) (11/21/88)
Gilbert Cockton writes: >Intelligence arises through socialisation. So do other diseases, but that doesn't mean it's the only way to get sick. The Eskimos have lots of words to distinguish between various types of snow; would somebody from sci.lang care to give us a few more words for intelligence? Culturation is one contributing factor to our peculiar brand of intelligence. Our particular set of sensory detectors and motor actuators is another. Neurophysiological constraints and mechanisms is yet another. The chemical makeup of the earth is one. The physics of the universe plays a role. So!? Is our definition of intelligence really so limited as to exclude all other domains? Machines will exhibit the salient properties of human intelligence. A fun book to read is Braitenberg's "Vehicles: Experiments in Synthetic Psychology." Another is Edelman's "Neural Darwinism." Bury Descartes already. Connectionist modeling and neurobiological research will bear fruit; why fight it? We should already be starting on the other hard task: creating robust, self-organizing, self-motive, self-sustaining, self-replicating machines. Artificial Intelligence will look like a cake-walk compared to Artificial Life. Now, leave me alone while I wire-wrap this damn frog. -- Doug Salot || doug@feedme.UUCP || ...{zardoz,dhw68k}!feedme!doug "vox populi, vox canis"
throopw@xyzzy.UUCP (Wayne A. Throop) (11/22/88)
> peru@soleil.UUCP (Dave Peru) > Definition of Intelligence: > 1. Know how to solve problems. > 2. Know which problems are unsolvable. > 3. Know #1, #2, and #3 defines intelligence. > This is the correct definition of intelligence. If anyone disagrees, please > state so and why. I disagree, because this matches very poorly with what people seem operationally to mean when they use the term. Better matches are gotten by "Intelligence is the extent to which one knows things about stuff", or perhaps "Intelligence is the ability to come to know more things about more stuff." Of course, despite being better than Dave's attempts, my trial definitions suffer the same flaw as Dave's... namely, they don't tie down what it means to know something. And, in fact, this is really the basic question here, compared to which the question of "what is Intelligence" is a mere quibble, and that is "what does it mean to know something?". (And even more basic, and even harder, "what does it mean to mean something?".) It's questions like these that keep people arguing in talk.philosophy.misc. And we can see fairly clearly that the question about "to know" is what Dave is really getting at, because the assertions at the end of his posting: > If you take into account the unsolvable problems of Turing machines, > then this proves Artificial Intelligence is impossible. > "Artificial Intelligence" is an unsolvable problem. > Human beings are not machines. > Human beings are capable of knowing which problems are unsolvable, while > machines are not. ... boil down to the assertion that humans "know" things in some mysterious way different from the way that machines "know" things. If Dave wished to convince me of the assertions he made, he would have to convince me that machines and humans "know" things in ways fundamentally distinct (as opposed to being distinct only in complexity or superficial organization). -- Alan Turing thought about criteria to settle the question of whether machines can think, a question of which we now know that it is about as relevant as the question of whether submarines can swim. --- Edgser Dijkstra -- Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw
throopw@xyzzy.UUCP (Wayne A. Throop) (11/22/88)
> peru@soleil.UUCP (Dave Peru) > There is a misunderstanding what I meant by [...my definition of > Intelligence...] especially [...the part reading "know which > problems are unsolvable." ...] > Human beings KNOW the "halting problem for Turing machines", my point is > that machines can NOT know the "halting problem for Turing machines". But there is no single "halting problem for Turing machines". There is only the halting problem for a particular machine programed in a particular way. Other "machines" can "know" the problems of some given machine. > Please describe how you would give this knowledge to a computer. If by "this knowledge", is meant the ability to solve the halting problem for all possible machines, it is of course impossible. But then again, there is no particular reason to think that humans can solve the problem for all machines either, as pointed out in Hofstadter's rebuttal to Lucas in "The Mind's I". The rest of Dave's posting boils down to the notion that > Finite machines cannot understand "infinity". ... which seems an interesting, though unsupportable assertion. All that is needed to model/understand "infinity" is the notion of boundlessness. Finite machines can encompas this notion easily, as in symbolic mathematical systems dealing with infinite series, and prolog and like systems having the ability to specify "infinite" lists or sequences. Further, is there any reason to suppose that humans are "infinite" in any relevant respect? I see no particular reason to suppose so. -- #2: ... just think of what we'll have! #6 A row of cabbages. #2 But highly educated cabbages! --- from "The Prisoner" -- Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw
smryan@garth.UUCP (Steven Ryan) (11/22/88)
>/ >toward human beings as being machines. What do you mean by machine? I mean a turing machine or any proven equivalent system. -- -- s m ryan +---------------------------------------------------------------+--------------+ | And they looked across the PDP7 and saw that it was idle and | OSF is the | | without users. They `said let there be code' and it was. | antiUnix. | +---------------------------------------------------------------+--------------+ There was a read and a write and it was the first memory cycle.
smryan@garth.UUCP (Steven Ryan) (11/22/88)
>" Just a short note that is a mortal sin for a mathematician to say >" `arb***ry.' Penance usually is usually five Hail Davids. >" >" The word `infinite' only occurs before `set,' `infinite set,' where it has >" a very precise meaning. >" >" The correct term is `arbitrary.' > >Now it makes even more sense. Without wanting to start a jihad, but in most math and computer science, there is a real attempt to avoid the word `infinity' and `infinite' really is restricted to `infinite set.' That little thing that looks like a propellor is read as `unbounded': x -> oo, f(x) -> oo. `As x increases unbounded, f at x increases unbounded.' The problem is that `infinity' is such a transcendental kind of word, that any attempt to define it causes problem. Aleph-0 is an infinite set, but 2**aleph-0=aleph-1 is an even larger infinite set. If we try to define infinity as the largest possible set, then we are left with infinity=2**infinity. `Arbitrary' means the same, more or less, as `unbounded.' Finally, note that `unbounded' does not always mean infinite. If a Turing machine accepts, its tape is unbounded, but it must be finite. -- -- s m ryan +---------------------------------------------------------------+--------------+ | And they looked across the PDP7 and saw that it was idle and | OSF is the | | without users. They `said let there be code' and it was. | antiUnix. | +---------------------------------------------------------------+--------------+ There was a read and a write and it was the first memory cycle.
sp299-ad@violet.berkeley.edu (Celso Alvarez) (11/22/88)
In article <151@feedme.UUCP> doug@feedme.UUCP (Doug Salot) asks: >would somebody from sci.lang care to give us a few more >words for intelligence? Intelligence is the art of sounding intelligent. Celso Alvarez sp299-ad@violet.berkeley.edu
ok@quintus.uucp (Richard A. O'Keefe) (11/22/88)
In article <151@feedme.UUCP> doug@feedme.UUCP (Doug Salot) writes: >Machines will exhibit the salient properties of human intelligence. >A fun book to read is Braitenberg's "Vehicles: Experiments in >Synthetic Psychology." Another is Edelman's "Neural Darwinism." >Bury Descartes already. Connectionist modeling and neurobiological >research will bear fruit; why fight it? To continue this rather constructive approach of suggesting good books to read that bear on the subject, may I recommend Women, Fire, and Dangerous Things -- what categories reveal about the mind George Lakoff, 1987 U of Chicago Press, ISBN 0-226-46803-8 I don't think the data he presents are quite as much of a challenge to the traditional view of what a category is as he thinks, provided you think of the traditional view as an attempt to characterise ``valid'' categories rather than actual cognition, just as classical logic is an attempt to characterise valid arguments rather than what people actually do. As an account of what people do, it is of very great interest for both AI camps, and it provides *evidence* for the proposition that non-human intelligences will not "naturally" use the same categories as humans. As for connectionist modelling, it doesn't tell us one teeny tiny little thing about the issues that Lakoff discusses that "classical AI" didn't. Why pretend otherwise? Case-based reasoning, now, ...
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (11/22/88)
In article <1791@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >In article <1908@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: >> >>Intelligence arises through socialisation. >> >Why is this a good argument against the possibility of machine intelligence? Cos you can't take a computer, not even the just truly awesomest nooral network ever, to see the ducks, get it to throw them bread, etc, etc. Take a walk through your life. Can you really see a machine going through that with an identical outcome? If so, lay off the cyberpunk and get some fresh air with some good folk :-) -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert
leverich@randvax.UUCP (Brian Leverich) (11/23/88)
In article <17@ubc-cs.UUCP> morrison@grads.cs.ubc.ca (Rick Morrison) writes: > >Does anyone in this group actually _do_ AI? > Yup, there are readers who are researchers in areas generally considered to be AI-related disciplines. Quality of the postings seems to have fallen off as quantity has climbed. Perhaps we need a talk.ai newsgroup? :-) -B -- "Simulate it in ROSS" Brian Leverich | U.S. Snail: 1700 Main St. ARPAnet: leverich@rand-unix | Santa Monica, CA 90406 UUCP/usenet: decvax!randvax!leverich | Ma Bell: (213) 393-0411 X7769
alexandr@surya.cad.mcc.com (Mark Alexandre) (11/23/88)
In article <17347@agate.BERKELEY.EDU> sp299-ad@violet.berkeley.edu (Celso Alvarez) writes: >In article <151@feedme.UUCP> doug@feedme.UUCP (Doug Salot) asks: >>would somebody from sci.lang care to give us a few more >>words for intelligence? > >Intelligence is the art of sounding intelligent. > And art is what I say it is.
bwk@mitre-bedford.ARPA (Barry W. Kort) (11/24/88)
I enjoyed reading Brad Brown's comments regarding Turing-computability and the limits of AI. It seems to me that one of the most powerful forms of human reasoning is reasoning by analogy, or model-based reasoning. We use models, metaphors, parables and analogies to transform problems from one domain to another, thereby borrowing ideas and translating them to novel situations. Reasoning by analogy requires pattern matching with respect to the structure of a knowledge base. We look at the shape of the semantic network or the shape of the tree and ask ourselves if we have encountered a similarly structured knowledge base before. Then, mutatis mutandis, we map out the analogy and fill in the missing pieces. Natural language is a powerful tool for suggesting metaphors. In mathematical circles, proof by analogy is emerging as an interesting research frontier. Saul Kripke has done seminal work in modal reasoning and intuitionist logic, which formalize these ideas. I wonder how the "limits of AI" argument will look after machines learn how to mimic human-style model-based reasoning. --Barry Kort
smryan@garth.UUCP (Steven Ryan) (11/24/88)
I've heard mazerunners define intelligence as the ability to learn. Place
motivated rats, dogs, humans, .... in a maze and see how many trials it
takes for them to learn the maze.
Well, this isn't math, so maybe the Pope will be quiet.
--
-- s m ryan
--------------------------------------------------------------------------------
As loners, Ramdoves are ineffective in making intelligent decisions, but in
groups or wings or squadrons or whatever term is used, they respond with an
esprit de corps, precision, and, above all, a ruthlessness...not hatefulness,
that implies a wide ranging emotional pattern, just a blind, unemotional
devotion to doing the job.....linhart@topaz.rutgers.edu (Phil) (11/24/88)
The gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes (jokingly?): [ How does "Intelligence arises thru socialisation." exclude machines? ] -=> Cos you can't take a computer, not even the just truly awesomest -=> nooral network ever, to see the ducks, get it to throw them bread, -=> etc, etc. Nor a blind paraplegic. If a computer can launch a missile, then surely it can launch a crust of bread. Why, it'd be like shooting ducks in a pond... :-) -=> Take a walk through your life. Can you really see a machine going -=> through that with an identical outcome? If so, lay off the cyberpunk -=> and get some fresh air with some good folk :-) Why need it be identical? Human intelligence is the example, not the definition. ( The smiley may have applied to the whole message, seeing as how the poster works in a CS department. Think before you flame. )
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (11/25/88)
In article <1918@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: >Cos you can't take a computer, not even the just truly awesomest >nooral network ever, to see the ducks, get it to throw them bread, >etc, etc. > >Take a walk through your life. Can you really see a machine going >through that with an identical outcome? If so, lay off the cyberpunk >and get some fresh air with some good folk :-) > I agree that we aren't nearly advanced enough to see our way through to making a robot capable of "acting human", but why do you think that this is impossible? (For now let's not assume I am saying that we SHOULD try to make a mechanical human, that's another question.) Did we not become what we are through a natural and (at least potentially) comprehensible process? Then could we not be functionally duplicated, given sufficient knowledge? It isn't like we were talking about exceeding the speed of light or making time run backwards. Or is it? I wish I could get you to say why you really don't believe it is possible. Or do you think we are just dumb American "techies" who wouldn't be able to understand your learned discourse? (If Decartes answered this to your satisfactorily, pray give us the reference, then.)
maddoxt@novavax.UUCP (Thomas Maddox) (11/25/88)
In article <1791@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >In article <1908@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: >>Intelligence arises through socialisation. >Why is this a good argument against the possibility of machine intelligence? >If a large enough neural network could be created, could it also not >receive socialization a la HAL? Not in Cockton's eyes. He wants to keep the argument focused on what he calls "socialization" rather than experience, which is the central issue. Almost certainly, whatever we mean when we say intelligence has reference to abilities acquired through interaction with the universe. However, if like Cockton you restrict the possibilities of acquisition of intelligence to social situations, then you have demonstrated (albeit through circular reasoning) the impossibility of machine intelligence. Which is of course the hobbyhorse he continues to ride.
jack@cs.glasgow.ac.uk (Jack Campin) (11/25/88)
bwk@mbunix (Kort) wrote: > In mathematical circles, proof by analogy is emerging as an interesting > research frontier. Saul Kripke has done seminal work in modal reasoning > and intuitionist logic, which formalize these ideas. I find this utterly unintelligible. What has intuitionistic logic got to do with analogy? What specific work of Kripke's are you talking about? [ To anticipate one likely question: if anyone wants to know what modal and intuitionistic logics are and what Kripke's contributed to them, read the relevant articles in the Handbook of Philosophical Logic. I am not very interested in debating people who haven't read some exposition like that. ] I am not saying there can't be a logic of analogy - though I have no idea what shape it would take. -- ARPA: jack%cs.glasgow.ac.uk@nss.cs.ucl.ac.uk USENET: jack@cs.glasgow.uucp JANET:jack@uk.ac.glasgow.cs useBANGnet: ...mcvax!ukc!cs.glasgow.ac.uk!jack Mail: Jack Campin, Computing Science Dept., Glasgow Univ., 17 Lilybank Gardens, Glasgow G12 8QQ, SCOTLAND work 041 339 8855 x 6045; home 041 556 1878
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (11/26/88)
In article <819@novavax.UUCP> maddoxt@novavax.UUCP (Thomas Maddox) writes: >However, if like Cockton you restrict the >possibilities of acquisition of intelligence to social situations, Intelligence is a social construct. The meaning of the word is defined through interaction. Dictionary definitions are irrelevant, and certainly never accurate or convincing. I have keep referring to the arguments against (or grounds for failure of) the C18 encyclopadists. Dictionaries arose in the enlightenment as well. Diderot, amongst others, recognised that the impossibility of prescribing meaning was a major obstacle, if not an unavoidable barrier, to the encyclopaedic endeavour. If dictionaries were only meant as spelling aids, there was less of a problem here. Since AI is just Diderot on disc, arguments against the C18 encyclopaedists, apart from being more convincing than the encyclopaedists' case, are also highly relevant today. Someone mailed me with the ignorant comment that C18 philosophy was adolescent and whiggishly outdated by modern developments. Is it hell. Read before you wallow in ignorance. Wittgenstein however backs up much of the case against the encyclopaedists. His arguments on the centrality of practice to knowledge and meaning rule out a logocentric theory of truth. I regard all symbol systems as effectively logocentric. Intelligence can only be acquired in social situations, since its presence is only acknowledged in social situations. The meanings are fluid, and will only be accepted (or contested) by humans in social contexts. AI folk can do what they want, but no one will ever buy their distortions, nor can they ever have any grounds for convincement in this case. What I am saying is that you cannot prove anything in this case by writing programs. Unlike sociology, they are irrelevant. You can stick any process you like in a computer, but its intelligence is a a matter for human negotiation. It may seem smart at first, but after a few days use in a real task? (ho, ho, ho - ever seen that with an AI program? - Prospector does seem to have done well, so no quibbles here. Anything else though? Also, even Prospector's domain restricted, unlike smart humans. AI cannot prove anything here. It can try to convince (but doesn't because of a plague of mutes), but the judgement is with the wider public, not the self-satisfied insiders. Now brave freedom fighter against the tyranny of hobbyhorses, show me my circular reasoning? linhart@topaz.rutgers.edu (Phil) writes in <Nov.24.09.07.42.1988.6716@topaz.rutgers.edu> > The gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes (jokingly?): > -=> Cos you can't take a computer, not even the just truly awesomest > -=> nooral network ever, to see the ducks, get it to throw them bread, > -=> etc, etc. Not completely jokingly, just a less direct European style. There was a serious point in there, but the etc. etc. marked out my feeling that full elaboration was unnecessary. I'll try to sum up in a more direct manner for SAT level literacy :-) (but really, it's a question of styles across cultures, which is ironic, for those cultures which understand irony that is!) Until a machine can share in socialisation, as a normal human, it will not be able to match the best human capabilities for any task. Thus > a blind paraplegic. does suffer some disadvantages, but for reasons which I cannot comprehend, but wonder at, such an individual can still get a lot out of life. This presence of humanity in the face of gross disability (or in the face of cruel oppression, e.g. U.S. slavery), is, for me, further proof that a mechanistic or biological account of being is going to miss out on the fundamentals of being. I'd still take a blind paraplegic to see the ducks. Even though they couldn't see, they could hear. Even though they can't throw, they might drop. If not, I'd throw for them. > If a computer can launch a missile, then surely it can launch a crust > of bread. Why, it'd be like shooting ducks in a pond... :-) And I don't think for one minute your machine you reflect on the morality of its action, as a group of children would. (no :-)) > Human intelligence is the example, not the definition. Example for what? I need to see more of the argument, but this already looks a healthier position than some in AI. > ( The smiley may have applied to the whole message, seeing as how the > poster works in a CS department. Think before you flame. ) This is Britain. You will find a range of people working in CS departments. As part of the Alvey programme, HCI research was expanded in the UK. You'll find sociologists, historians, fine artists, literature graduates, philosophers and educationalists working in CS departments here, as well as psychologists and ergonomists. As part of the Alvey HCI programme, technical specialists HAVE come in (perhaps unfairly at times) for a lot of flack over the way they design (on a good day) computer systems. No need to think before I flame, as we don't expect blind dormitory brotherhood loyalty over here. This is a university, not a regiment. HCI isn't about automating everything (the AI mania), it's about improved use of computers, whatever the level of technology. The CS contribution is on the construction of programs. Robust and verifiable programs are a major requirement for improved interaction, but this alone will not significantly stop the problems of misuse, disuse and abuse associated with poorly designed human-computer systems. Good code bricklayers and engineers will design good walls, but they will rarely make it to a habitable building, and even then never by intent. I talk of roles here of course, not individuals. Some good CS trained code brickies are also sensitive to human issues, and effectively so. Many of the non-CS graduates in UK HCI research also happen to be good code brickies as well. To build habitable systems, you need technical *AND* human skills. There are two roles, and you can fill them with one or many people. Both roles MUST be filled though. AI rarely fills either. CS needs taking down a peg on some issues (largely ones it ignores when it cannot, and thus does not), so there's no worry about righteous indignation from the unbigoted. -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (11/26/88)
In article <819@novavax.UUCP> maddoxt@novavax.UUCP (Thomas Maddox) writes: >>receive socialization a la HAL? > > Not in Cockton's eyes. He wants to keep the argument focused >on what he calls "socialization" rather than experience, which is the >central issue. Almost certainly, whatever we mean when we say >intelligence has reference to abilities acquired through interaction >with the universe. However, if like Cockton you restrict the >possibilities of acquisition of intelligence to social situations, >then you have demonstrated (albeit through circular reasoning) the >impossibility of machine intelligence. Which is of course the >hobbyhorse he continues to ride. Oh, I see. I thought he was going to trot out the old self-reference argument or some rehash of Dreyfus. That is interesting. But that would mean that we couldn't really talk about intelligence as being a property of an individual, but collectively of a species, sort of like a hive of bees or something. This would rule out intelligence, I suppose, in species without significant socializing, such as egg-layers where the parents don't stick around. I suppose one could speculate on two types of attempts to create artificial intelligence(s). One would be to create an artificial member of the human species (like an android) and another to create an entire artificial species which the individuals would interact with each other. The latter seems more difficult and probably more dangerous. I don't think Cockton has given a good argument for why eventually (given enough understanding of neural connections in humans and sufficient advances in silicon) an artificial child could not be created which could be socialized much as our real children are (even as far as teaching it about feeding ducks).
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (11/28/88)
In article <1976@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: > > Intelligence can only be acquired in social situations, since its > presence is only acknowledged in social situations. If a tree falls in the woods and nothing with ears is around, does it make a sound? I think it is imaginable that there could be in this universe solitary entities that "think". The social definition of intelligence might also lead to making the mistake of attributing intelligence to natural processes which (at least by current definitions) are not, e.g. theories that planets hold their orbits voluntarily in response to God's law, which were at one time common. > AI cannot prove anything here. It can try to convince (but doesn't > because of a plague of mutes), but the judgement is with the wider > public, not the self-satisfied insiders. > AI will not convince except with results. The proof of the pudding is in the eating. So far there has been too much talking and too little results. As for the value of HCI endeavors, the same applies there. We'll see how much the eclectic nature of this effort pays off by the hard results, won't we? > >Not completely jokingly, just a less direct European style. .. >I'll try to sum up in a more direct manner for SAT level literacy :-) >(but really, it's a question of styles across cultures, which is >ironic, for those cultures which understand irony that is!) .. >in the face of cruel oppression, e.g. U.S. slavery), Your not so subtle needling of Americans would seem less hypocritical if it originated in a country (and continent) whose own hands were cleaner. >further proof that a mechanistic or biological account of being is >going to miss out on the fundamentals of being. > I agree that socialization of the machine is probably going to be needed to create anything that we might recognize as an artificial person. But can you talk about what the fundamentals of being are and what denies them to even the most complex artifact (other than feeding the ducks)? Is this some supernatural notion, or is it the result of evolution, or culture, or some other natural process? >> Why, it'd be like shooting ducks in a pond... :-) >And I don't think for one minute your machine you reflect on the >morality of its action, as a group of children would. (no :-)) > Depends entirely what the children have been taught. There are lots of children who wouldn't reflect on the morality of such actions. Conversely, the program could well be made to reflect on such moral issues.
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (11/28/88)
From article <1976@crete.cs.glasgow.ac.uk>, by gilbert@cs.glasgow.ac.uk (Gilbert Cockton): " .... I regard all symbol systems as effectively logocentric. Would you tell us why *we* should? E.g., taking a look at some axiom sets for symbolic logic, it is less than obvious that prescribing the meanings of words is what is going on. " Intelligence can only be acquired in social situations, since its " presence is only acknowledged in social situations. whoops. " The meanings " are fluid, and will only be accepted (or contested) by humans in " social contexts. AI folk can do what they want, but no one will " ever buy their distortions, nor can they ever have any grounds for " convincement in this case. As applied to the meaning of 'intelligence', there seems to be a kind of contradiction here. "Noone will ever buy their distortions" translates to "language use ascribing intelligence to computers will never come to be regarded as fair play in the language game." If we accept the translation, we see that the thesis is obviously false, since language has already come to be used this way. Listen around. "Dumb program!" "If you put the name in quotes, the computer will understand you." ... Greg, lee@uhccux.uhcc.hawaii.edu
sbigham@dukeac.UUCP (Scott Bigham) (11/28/88)
In article <1972@garth.UUCP> smryan@garth.UUCP (Steven Ryan) writes: >I've heard mazerunners define intelligence as the ability to learn. Place >motivated rats, dogs, humans, .... in a maze and see how many trials it >takes for them to learn the maze. Just one problem. The computer you're reading this article on can learn the maze in -one- trial, and I don't think anyone would call it intelligent. sbigham -- Scott Bigham "The opinions expressed above are Internet sbigham@dukeac.ac.duke.edu (c) 1988 Hacker Ltd. and cannot be USENET sbigham@dukeac.UUCP copied or distributed without a ...!mcnc!ecsgate!dukeac!sbigham Darn Good Reason."
engelson@cs.yale.edu (Sean Philip Engelson) (11/28/88)
In article <1918@crete.cs.glasgow.ac.uk>, gilbert@cs (Gilbert Cockton) writes: >In article <1791@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >>In article <1908@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: >>> >>>Intelligence arises through socialisation. >>> >>Why is this a good argument against the possibility of machine intelligence? >Cos you can't take a computer, not even the just truly awesomest >nooral network ever, to see the ducks, get it to throw them bread, >etc, etc. Yet, my good friend. YET. >Take a walk through your life. Can you really see a machine going >through that with an identical outcome? Of course not. Intelligent machines won't act much like humans at all. They will have different needs, different feelings, different goals, plans, desires for life than we. But they'll be no less intelligent, thinking, feeling beings than we, for it. >If so, lay off the cyberpunk >and get some fresh air with some good folk :-) Perhaps you should lay off the mysticism and get some fresh rationality with some good folk :-) >-- >Gilbert Cockton, Department of Computing Science, The University, Glasgow > gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert Sean ---------------------------------------------------------------------- Sean Philip Engelson, Gradual Student Yale Department of Computer Science 51 Prospect St. New Haven, CT 06520 ---------------------------------------------------------------------- The frame problem and the problem of formalizing our intuiutions about inductive relevance are, in every important respect, the same thing. It is just as well, perhaps, that people working on the frame problem in AI are unaware that this is so. One imagines the expression of horror that flickers across their CRT-illuminated faces as the awful facts sink in. What could they do but "down-tool" and become philosophers? One feels for them. Just think of the cut in pay! -- Jerry Fodor (Modules, Frames, Fridgeons, Sleeping Dogs, and the Music of the Spheres)
rjc@aipna.ed.ac.uk (Richard Caley) (11/28/88)
In article <1976@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: > Intelligence is a social construct. The meaning of the word is > defined through interaction. Dictionary definitions are > irrelevant, and certainly never accurate or convincing. This is one argument. . . > Intelligence can only be acquired in social situations, since its > presence is only acknowledged in social situations. This is another. They are not in any way equivalent nor does one necessarilly follow from the other. ( I agree with the first but not with the second - the property of being a chair is also socially defined, however a naturally formed object can be recognised as a chair without having attained its form and function via social interaction ). > The meanings > are fluid, and will only be accepted (or contested) by humans in > social contexts. AI folk can do what they want, but no one will > ever buy their distortions, nor can they ever have any grounds for > convincement in this case. So what's new here. I should think you would have to look hard for an AI researcher who didn't believe this. This is what the Turing test is all about, putting a machine in a context where its being a machine will not bias social interaction and seeing if it accepted as intelligent. What distortions are you talking about here? This sounds like a straw man to me. > What I am saying is that you cannot prove anything in this case by > writing programs. Unlike sociology, they are irrelevant. Now you argue against yourself. If intelligence can only be recognised via social interaction then the systems which are puported to have this property _must_ be built ( or programmed ) to be tested. Sociology can not say yes or no, though it can point out hopeless paths. You have yourself, if I remember correctly, said that AI workers lack training in experimental design as would be given to psycology undergraduates - are you now saying that experimentation is useless after all? > Also, even Prospector's > domain restricted, unlike smart humans. Most "smart humans" also have restricted domains ( though admittedly rather larger than that of Prospector ). I doubt many people have expert level knowledge in, say, both 12th century history and particle physics. Where people differ from so called "expert systems" is in their ability to cope with non "expert" tasks, such as throwing bread to ducks. > Now brave freedom fighter against the tyranny of hobbyhorses, show > me my circular reasoning? I am not the brave freedom fighter adressed, but . . . . The argument seems to go something like the following - 1 ) "Intelligence" can only be judged by social interaction with the supposedly intelligen system. 2 ) I can not concieve of a computer system capable of succesfully interacting in this way. 3 ) Therfore no computer system can ever be intelligent. 4 ) Therfore (2) Just saying that intelligence requires socialisation does not prove the impossibility of machine intelligence without the lemma that machines can not be social entities, which is at least as big an assumption as the impossibility of intelligence. >Until a machine can share in socialisation, as a normal human, it will >not be able to match the best human capabilities for any task. I agree with reservations ( there are tasks in which a machine can exceed the capabilities of any human, take leveling a city as an example.) >And I don't think for one minute your machine you reflect on the >morality of its action, as a group of children would. (no :-)) This would seem to be based on another circular argument - machines can not be socialised, so machines cannot acquire a morallity, so I would never accept a machine as a social entity . . . >> Human intelligence is the example, not the definition. >Example for what? I need to see more of the argument, but this >already looks a healthier position than some in AI. If I may once again answer for the person adressed ( I have managed to delete their name, my appologies ), I believe he meant an example for the kind of abilities and behaviours which are the target for AI. That is, human beings are intelligent entities, but the reverse is not neccesarilly the case. >HCI isn't about automating everything (the AI mania), Except in a derivative sence, AI is not about automation. Although it often procedes by trying to automate some task, in order to gain insight into it, that is a research strategy, not a definition of the field. > { paragraph about system design vs. implementation } > >Both roles MUST be >filled though. AI rarely fills either. So what, AI is not doing HCI, it is also not doing biochemistry, why should it be? AI has created some tools which people are using to create computer systems for various tasks, and you are quite at liberty to critisise the design of such systems. However that is not critisism of AI any more than critisism of the design of a television is critisism of the physics which lead to the design of the electronics. -- rjc@uk.ac.ed.aipna AKA rjc%uk.ac.ed.aipna@nss.cs.ucl.ac.uk "We must retain the ability to strike deep into the heart of Edinburgh" - MoD
maddoxt@novavax.UUCP (Thomas Maddox) (11/28/88)
In article <1976@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: >In article <819@novavax.UUCP> maddoxt@novavax.UUCP (Thomas Maddox) writes: >>However, if like Cockton you restrict the >>possibilities of acquisition of intelligence to social situations, > > Intelligence is a social construct. The meaning of the word is > defined through interaction. Dictionary definitions are > irrelevant, and certainly never accurate or convincing. Indeed dictionary definitions are irrelevant to this argument, and I have no idea why Cockton brought them in. In fact, *definition* is not in question; *acquisition* is. > I have keep referring to the arguments against (or grounds for failure of) > the C18 encyclopadists. Dictionaries arose in the enlightenment as > well. More irrelevancies. Prescribing meaning is no more an issue than definition was to begin with. > Since AI is just Diderot on disc, arguments against the C18 > encyclopaedists, apart from being more convincing than the > encyclopaedists' case, are also highly relevant today. Someone > mailed me with the ignorant comment that C18 philosophy was adolescent > and whiggishly outdated by modern developments. Is it hell. Read > before you wallow in ignorance. Wittgenstein however backs up much > of the case against the encyclopaedists. His arguments on the > centrality of practice to knowledge and meaning rule out a > logocentric theory of truth. I regard all symbol systems as > effectively logocentric. "AI is just Diderot on disk": a remarkable statement, one that if true would surprise a number of people; if Diderot could be around, him most of all. In what sense is your statement true? one might ask. Also, are you referring to the Diderot of the Encyclopedia, of Rameau's Nephew, Jacques the Fatalist? Then Wittgenstein gets dragged in, and Derrida is invoked indirectly through the mention of logocentrism. This is not an argument so much as a series of irrelevancies and non sequiturs. > Intelligence can only be acquired in social situations, since its > presence is only acknowledged in social situations. The meanings > are fluid, and will only be accepted (or contested) by humans in > social contexts. AI folk can do what they want, but no one will > ever buy their distortions, nor can they ever have any grounds for > convincement in this case. Again a series of disconnected and entirely unsupported remarks. Given that intelligence is acknowledged in social situations, how does this affect the case for or against AI? Presumably any artificial intelligence could be acknowledged in a social situation as easily as organic intelligence howsoever defined. > What I am saying is that you cannot prove anything in this case by > writing programs. Unlike sociology, they are irrelevant. You can > stick any process you like in a computer, but its intelligence is a > a matter for human negotiation. This is no more than a restatement of Turing's position, hardly, therefore, a refutation of AI. [. . .] > AI cannot prove anything here. It can try to convince (but doesn't > because of a plague of mutes), but the judgement is with the wider > public, not the self-satisfied insiders. > > Now brave freedom fighter against the tyranny of hobbyhorses, show > me my circular reasoning? At this point your reasoning is not so much circular as non-existent. You presume that the current non-existence of artificial intelligence proves the impossibility of same. I leave to the reader the elementary disproof of this position. >linhart@topaz.rutgers.edu (Phil) writes in <Nov.24.09.07.42.1988.6716@topaz.rutgers.edu> > >> The gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes (jokingly?): >> -=> Cos you can't take a computer, not even the just truly awesomest >> -=> nooral network ever, to see the ducks, get it to throw them bread, >> -=> etc, etc. > >Not completely jokingly, just a less direct European style. There was >a serious point in there, but the etc. etc. marked out my feeling that >full elaboration was unnecessary. Note Cockton's implication that those wily Europeans need to be explained to us simpletons over here in the USA. >I'll try to sum up in a more direct manner for SAT level literacy :-) Sorry, pal, but that smiley doesn't obscure the offensiveness of the remark. Your continuing snottiness in these matters perhaps deserves equally snotty and waspish rebuttal. >(but really, it's a question of styles across cultures, which is >ironic, for those cultures which understand irony that is!) Now what cultures might those be, Cockton? Could you be referring to all those complex and indirect European cultures, where everyone could be expected to sift through the fine ironies you employ to the wisdom buried within? Certainly you couldn't be referring to American culture, that callow, mechanistic, unironic haven for soulless technocrats. You really are a pretentious, overbearing shit. While you are certainly free to have contempt for things American, you are not free to substitute your prejudices for discussion without being answered with a rudeness equal to your own. [. . .] >This is Britain. You will find a range of people working in CS >departments. As part of the Alvey programme, HCI research was >expanded in the UK. You'll find sociologists, historians, fine >artists, literature graduates, philosophers and educationalists working >in CS departments here, as well as psychologists and ergonomists. >As part of the Alvey HCI programme, technical specialists HAVE come in >(perhaps unfairly at times) for a lot of flack over the way they >design (on a good day) computer systems. No need to think before I >flame, as we don't expect blind dormitory brotherhood loyalty over >here. This is a university, not a regiment. Again the implication that *there* things are done properly, humanly, with a due respect for all human culture, while *here* . . . well, my dear, we are (see above) a narrow group of regimented technophiles. Cockton, I've said it before, but as you've done nothing to change my evaluation, I'll say it again: you read like a particularly narrowly-conceived language-generating program compounded of equal parts Dreyfus and Jeremy Rifkin. Now, however, you've apparently added a sub-program intended to reproduce the rude anti-Americanism of Evelyn Waugh on an especially nasty day.
ok@quintus.uucp (Richard A. O'Keefe) (11/28/88)
In Chapter 2 of "Ten Philosophical Mistakes", Mortimer J. Adler says We ordinarily speak of any living organism that has some consciousness of its environment and of itself as having a mind. We also attribute intelligence to that organism if, in addition to such consciousness, it reacts in some discriminating fashion to the environment of which it is aware. It should be added, perhaps, that we generally regard mind and intelligence as the means by which sentient organisms learn from experience and modify their behaviour accordingly. By these criteria, the only animals to which we would not attribute mind or intellignece are those the behaviour of which is completely determined by innate, preformed patterns of behaviour that we call instincts. ... For [humans] as well as for [animals], mind or intelligence stands for faculties or powers employed in learning from experience and in modifying behaviour in consequence of such learning. This definition of intelligence would appear to be one that could meaningfully be applied to machines. A machine which learned in fewer trials, or was capable of learning more complex ideas, could be said to possess super-human intelligence. It is interesting that Adler's definition implicitly takes into account the varying sensory capacities of organisms: unlike a fish we cannot learn to adapt our behaviour to the presence or absence of a weak electric field, but that is not a defect of intelligence, because we are incapable of experiencing the presence or absence of the field. It would be a defect of intelligence to be unable to learn from something that we _are_ capable of experiencing. In fact, psychologists sometimes test whether an organism can learn to dsicriminate between two conditions as a method of determining whether the organism can perceive the difference.
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (11/28/88)
In article <1816@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >it make a sound? I think it is imaginable that there could be >in this universe solitary entities that "think". Not the same as intelligence. You can think dumb things. >AI will not convince except with results. Wrong, and you misunderstand the problem of 'results' in human science. An experiment proves very little in psychology. It is the design and the generalisation which are crucial, and these can only proceed by argument. >Your not so subtle needling of Americans would seem less >hypocritical if it originated in a country (and continent) Before this gets out of hand, the reference to U.S. slavery was just the best example I know of, due to the quality of the historical sources. I wasn't singling out America, and I apologise to anyone who was (mistakenly) annoyed, either by my posting, or by believing this misinterpretation. As for cultural styles, I was only drawing attention to the differences. I was not trying to create division or claim supriority. As for approaches to literacy, well there are differences here, and it looks like at least one American takes gentle jokey references to this the wrong way. Still, this is an international net, and we can't all tailor our style of writing to suit one culture. >Is this some supernatural notion, or is >it the result of evolution, or culture, or some other natural process? Don't know, do you? There is a whole range of experience which does not seem to hav a mechanical basis. Which behaviour is AI trying to cover (and do say 'intelligent' behaviour, since this means nothing here)? -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (11/28/88)
In article <2717@uhccux.uhcc.hawaii.edu> lee@uhccux.uhcc.hawaii.edu (Greg Lee) writes: >Would you tell us why *we* should? E.g., taking a look at some axiom >sets for symbolic logic, it is less than obvious that prescribing >the meanings of words is what is going on. On the contrary, most AIers believe the assertion that logic encapulates the rules of thought, and that all sentences can be given a semantics in formal logic (note how some famous mathematical logicians disagree and stick to formal languages as being very different things). >translates to "language use ascribing intelligence to computers >will never come to be regarded as fair play in the language game." >If we accept the translation, we see that the thesis is obviously >false, since language has already come to be used this way. Listen >around. "Dumb program!" "If you put the name in quotes, the >computer will understand you." ... None of your examples would be accepted as anything except sloppy language, but acceptable sloppiness given that "Dumb programmer who wrote this program" or "Dumb design team who got the functionality of this software wrong" are far too long winded. Part of the consciousness raising of HCI is that programs are never dumb, only the programmers and designers who make them inadequate, or the design which commercial considerations forced on the designers or programmers. Anyone who talks of computers "understanding" does so: a) to patronise users whom they don't know how to instruct properly; b) because they are AI types. The majority of competent instructtors and realists wouldn't use "understand" for "fit the dumbly designed syntax". People are loose with their language. What counts is what they stick out for. -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (11/28/88)
In article <821@novavax.UUCP> maddoxt@novavax.UUCP (Thomas Maddox) writes: > > Cockton, I've said it before, but as you've done nothing to >change my evaluation, I'll say it again: you read like a particularly >narrowly-conceived language-generating program compounded of equal >parts Dreyfus and Jeremy Rifkin. Now, however, you've apparently added >a sub-program intended to reproduce the rude anti-Americanism of Evelyn >Waugh on an especially nasty day. Well said! Could this display of snobbery reflect an attempt at creating a simulation of a hierarch of the class-bound British educational establishment? Maybe he dislikes our meritocracy, but at least here a child of working class parents can become a professional without having to learn to disguise an accent.
jackson@freyja.css.gov (Jerry Jackson) (11/29/88)
In article <44150@yale-celray.yale.UUCP>, engelson@cs (Sean Philip Engelson) writes: > >Of course not. Intelligent machines won't act much like humans at >all. They will have different needs, different feelings, different >goals, plans, desires for life than we. But they'll be no less >intelligent, thinking, feeling beings than we, for it. I can accept the needs, goals, and plans... but why does everyone assume that an intelligent machine would be a *feeling* being? I see no reason to assume that an IM would be capable of experiencing anything at all. This doesn't imply that it wouldn't be intelligent. For instance: some machines are already capable of distinguishing blue light from red. This doesn't mean that they have anything like our *experience* of blue. (Or pain, or sorrow, or pleasure... etc.) Personally, I think this is a good thing. I would rather not have a machine that I would be afraid to turn off for fear of harming *someone*. It does seem that our experience is rooted in some kind of electro-chemical phenomenon, but I think it is an incredible leap of faith to assume that logic circuits are all that is required :-). BTW: It is perfectly consistent to assume that only I experience anything. i.e. it seems other people can be explained quite well without resorting to notions of experience. I claim that it is very likely that this position would be accurate in the case of intelligent machines (at least... intelligent digital computers ;-) --Jerry Jackson
sarge@metapsy.UUCP (Sarge Gerbode) (11/29/88)
In article <757@quintus.UUCP> ok@quintus.UUCP (Richard A. O'Keefe) writes: >In Chapter 2 of "Ten Philosophical Mistakes", Mortimer J. Adler says > > ... > For [humans] as well as for [animals], mind or intelligence > stands for faculties or powers employed in learning from > experience and in modifying behaviour in consequence of such > learning. > >This definition of intelligence would appear to be one that could >meaningfully be applied to machines. The significance of this definition would depend on what is to be included as "learning". A mere modification of behavior based on a change of environment would not, to my mind, qualify as "learning". For instance, the switching action of a thermostat in response to environmental changes in temperature would not entitle it to be considered to have "learned" anything, nor to be considered intelligent. And a person can exercise intelligence without behaving (physically), e.g. by thinking up a brilliant thought. Some very intelligent people ("effete intellectual snobs", I believe they used to be called :-) ) are very good at not applying their intelligence to real life. So the "behavior" part is extraneous to intelligence. It is the "learning" that is crucial. We could say that anything that could learn could be intelligent. Or, intelligence is the ability to learn. Intelligence tests were originally designed to predict school performance, i.e. learning ability, so that would fit this definition. The next question is whether machines could be said to "learn" in anything but a metaphorical sense. Perhaps they can be taught to behave in a way that imitates behavior that is thought to be consequent to actual learning, but would that mean that they actually "learned" anything? Each of us humans has direct subjective apperception of what it is to learn -- it is to acquire knowledge, to come to know something, to acquire a fact. What we do behaviorally with what we learn is another matter. Do machines have the same subjective experience that we do when we say we have learned something, or any subjective experience at all? It seems quite questionable. Since their behavior is completely explainable in terms of the hardware design, the software program, and the input data, Occam's Razor demands that we not attribute subjectivity to them. -- -------------------- Sarge Gerbode -- UUCP: pyramid!thirdi!metapsy!sarge Institute for Research in Metapsychology 950 Guinda St. Palo Alto, CA 94301
ok@quintus.uucp (Richard A. O'Keefe) (11/29/88)
In article <1821@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >Well said! Could this display of snobbery reflect an >attempt at creating a simulation of a hierarch >of the class-bound British educational establishment? >Maybe he dislikes our meritocracy, but at least here >a child of working class parents can become a professional >without having to learn to disguise an accent. Several false assumptions in here: (a) the USA has a meritocracy. A meritocracy is "rule by persons chosen for their superior talents or intellect". The ruling class in the USA is chosen for its ability to pay enough to look good in the media. As Ambrose Bierce put it: "The best Congress money can buy." [That may well be the best practical criterion, what do I know?] (b) maybe the assumption was that the educational system in the USA is meritocratic. In relation to students, this may well be so, but considering the number of students who have to go into debt to finance their education at tertiary level, and the growing home- schooling movement, one suspects "The best education money can buy." (c) A child of working class parents cannot become a professional in the UK without having to disguise an accent. Maybe I disbelieve this because I studied in Edinburgh, but I visited friends in Oxford where there were N different accents, _and_ working-class students. (d) This discussion is getting us anywhere. Once before I tried to give an account of why it was reasonable for people working on AI to pay little attention to sociology. This time I'm going to attempt a sociological explanation. It is a lot of work trying to stay informed in one subject, let alone several. I for one am trying to keep reasonably current in half a dozen topics, and I'm really stretched thin (my pockets are suffering too). I literally haven't got the _time_ to study the philosophy and sociology I would like to. (Adler and Bok are all I can manage at the moment, thank you.) So what do I do? I have to trust someone. Given the choice of trusting John McCarthy (say) or Gilbert Cockton (say), who do I trust? Well, one of these people belongs to some of the same fields that I do. If he puts me crook, a field _he_ helped found is injured. What's more, he keeps working on trying to solve the problems, and a few years ago came up with an approach which is of considerable mathematical interest, if nothing else. (I could say similar things about quite a lot of people in AI.) I claim that it is rational for me to trust McCarthy's (say) evaluation of the possibility of AI in preference to Cockton's. [It would _not_ be rational for me to prefer J.Random Hacker's; she hasn't the background of McCarthy.] There _is_ a very serious challenge to the possibility of AI (in the let's-build-a-god sense; the let's-make-amplifiers-for-the-mind camp can only gain) in Lakoff's "Women, Fire, and Dangerous Things". I think that's a very exciting book. He tackles some very basic topics about language, thought, and meaning, and attempts to show that the physical-symbol-system approach is founded on sand. But he doesn't leave us with mysteries like "socialisation"; an AI/NL person could expect to read this book and come away with some ideas to try. I would really like to see that book discussed in this newsgroup.
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (11/30/88)
In article <562@metapsy.UUCP> sarge@metapsy.UUCP (Sarge Gerbode) writes: > >We could say that anything that could learn could be intelligent. Or, >intelligence is the ability to learn. Intelligence tests were >originally designed to predict school performance, i.e. learning >ability, so that would fit this definition. > Then I presume that patients with various brain lesions, such as bilateral lesions of the hippocampus are to be considered non-intelligent? They certainly can't learn new facts (although there is good evidence that they still can be conditioned operantly to change their behavior). When presented with new problems to solve, they do about as well as normals, but on repeated presentation of the same problem, normals obviously improve quickly, whereas these patients do not. I would argue that their I.Q. is definitely not 0 by any reasonable definition of intelligence.
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (11/30/88)
From: ok@quintus.uucp (Richard A. O'Keefe) >In article <1821@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >> but at least here >>a child of working class parents can become a professional >>without having to learn to disguise an accent. > >(c) A child of working class parents cannot become a professional in the > UK without having to disguise an accent. Maybe I disbelieve this > because I studied in Edinburgh, but I visited friends in Oxford where > there were N different accents, _and_ working-class students. I agree that this probably isn't the newsgroup for this discussion, but I will just make this one explanation of what I said. I didn't mean this as a universal absolute about the UK but intended to mock the spirit of Cockton's stereotypes of the US. I also studied at Edinburgh and at Newcastle-upon-Tyne. My statement was based on my personal experiences in the UK. I met professionals from the working class who had changed their accent. I remember one chap that as he drank more beer at the pub his accent changed from "BBC" to Geordie. There was definitely the idea that a proper physician simply could not speak with a working class accent. That certainly is not true in the States, where some of the most prominent neurologists I know have strong New York City accents, for example. Others sound like the Southern sheriff in "Smokey and the Bandit". While a Scots accent in Edinburgh was fine, I wonder how it would play in London? I found British society to be an order of magnitude more class conscious than that of the US, and accent was the main way you were typed. This may have all changed since I was there (1976), but I doubt it. No one should get the idea that I don't like the British. I had a great experience and made a lot of friends, but all cultures have their faults, not just the US, ok?
ap1i+@andrew.cmu.edu (Andrew C. Plotkin) (11/30/88)
/>Of course not. Intelligent machines won't act much like humans at
/>all. They will have different needs, different feelings, different
/>goals, plans, desires for life than we. But they'll be no less
/>intelligent, thinking, feeling beings than we, for it.
/
/ I can accept the needs, goals, and plans... but why does everyone
/ assume that an intelligent machine would be a *feeling* being? I see
/ no reason to assume that an IM would be capable of experiencing
/ anything at all. This doesn't imply that it wouldn't be intelligent.
/ For instance: some machines are already capable of distinguishing blue
/ light from red. This doesn't mean that they have anything like our
/ *experience* of blue. (Or pain, or sorrow, or pleasure... etc.)
Aren't feeling and emotion automatic byproducts of thought? I find it hard to
imagine an entity capable of thought which would *not* have (for example) some
negative reaction to a threat, involving some attempt to rapidly find
alternatives, etc... which is to say, fear and excitement. Other emotional
responses can be argued similarly.
It's true that our physical bodies provide extra sensation by pumping out
adrenaline and so forth, but the original emotions are often generated by the
mind first and -then- relayed to the glands; for example, the panic you feel
when a pay cut notice appears on your desk.
(One can still argue that "emotional reactions" don't prove that the machine is
"really feeling emotion." I'll be glad to answer this point, if you can first
prove to me that *you* "really feel" pain, sorrow, or pleasure, and don't just
react mechanically...)
/ I would rather not have a
/ machine that I would be afraid to turn off for fear of harming
/ *someone*.
If you don't want a "someone", you'd better stay out of AI research... :-)
/ It does seem that our experience is rooted in some kind of
/ electro-chemical phenomenon, but I think it is an incredible leap of
/ faith to assume that logic circuits are all that is required :-).
It's a point of faith, certainly, since we don't have more than one example. But
I don't think it's unjustified, since nothing more than logic circuits has ever
been observed. ("Logic circuits" is a bit of a misnomer, since neurons don't act
like single standard logic gates. However, it looks possible to simulate them
with electronics.)
--Zjeff@lorrie.atmos.washington.edu (Jeff L. Bowden) (11/30/88)
In article <1985@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: >None of your examples would be accepted as anything except sloppy <language, but acceptable sloppiness given that "Dumb programmer who >wrote this program" or "Dumb design team who got the functionality of <this software wrong" are far too long winded. > <Part of the consciousness raising of HCI is that programs are never >dumb, only the programmers and designers who make them inadequate, or <the design which commercial considerations forced on the designers or >programmers. < >Anyone who talks of computers "understanding" does so: < > a) to patronise users whom they don't know how to instruct properly; < b) because they are AI types. > If someone says something to you and you don't understand is it a) Your fault? b) God's fault? c) Your mother's fault? d) The fault of some other thing to which you give credit (blame?) for your existence? Certainly it is the fault of the programmer if a program is deficient in understanding something, but it is certainly not sloppy English to say that the program does not understand. It doesn't. It was not imbued by its creator with the ability to understand. Fault has little to do with this. It appears to me that Mr. Cockton has an axe to grind with those who assume that every computer scientist accepts materialism.
bwk@mitre-bedford.ARPA (Barry W. Kort) (11/30/88)
In article <0XTukNy00Xol41W1Ui@andrew.cmu.edu> ap1i+@andrew.cmu.edu Andrew C. Plotkin writes: >I maintain that a human can be simulated by a Turing machine. Comments? As a human being, I occasionally cast lots to choose a course of action when my value system is balanced on the razor's edge between two alternatives. Because I want to be sure that my Random Number Generator is not deterministic (like a pseudo-random sequence generator), I use a quantum amplifier in my coin flipper. Correct me if I'm wrong. But a Turing Machine is obliged to follow a deterministic program. Hence a Turing machine cannot simulate my dice-tossing method. --Barry Kort
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (11/30/88)
From article <1985@crete.cs.glasgow.ac.uk>, by gilbert@cs.glasgow.ac.uk (Gilbert Cockton): " ... " People are loose with their language. What counts is what they stick " out for. I guess I didn't make myself clear. You had argued, as I understood you, that language is loose, and AI approaches do not take this adequately into account. But in judging the prospects for creating artificial minds, you use a preconceived notion of what intelligence "really" means, rather than letting the meaning emerge, loosely, from a social and conversational consensus (as appears to be happening). There appear to be two different and conflicting ideas about the nature of meaning in language involved. Greg, lee@uhccux.uhcc.hawaii.edu
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (12/01/88)
From article <562@metapsy.UUCP>, by sarge@metapsy.UUCP (Sarge Gerbode): " ... " Do machines have the same subjective experience that we do when we " say we have learned something, or any subjective experience at all? " It seems quite questionable. Since their behavior is completely " explainable in terms of the hardware design, the software program, " and the input data, Occam's Razor demands that we not attribute " subjectivity to them. A more proper application of Occam's Razor would be that it prevents us from assuming a difference between humans and machines in this regard without necessity. What does explaining behavior have to do with it? If I could explain your behavior, would this have the consequence that you cease to have subjective experience? Of course not. (If *you* could explain your behavior, perhaps the case could be made ...) Greg, lee@uhccux.uhcc.hawaii.edu
jackson@freyja.css.gov (Jerry Jackson) (12/01/88)
In article <IXYmnvy00XoGA0UVFk@andrew.cmu.edu>, ap1i+@andrew (Andrew C. Plotkin) writes: >/>goals, plans, desires for life than we. But they'll be no less >/>intelligent, thinking, feeling beings than we, for it. >/ >/ I can accept the needs, goals, and plans... but why does everyone >/ assume that an intelligent machine would be a *feeling* being? I see >/ no reason to assume that an IM would be capable of experiencing >/ anything at all. This doesn't imply that it wouldn't be intelligent. > >Aren't feeling and emotion automatic byproducts of thought? I find it hard to >imagine an entity capable of thought which would *not* have (for example) some >negative reaction to a threat, involving some attempt to rapidly find >alternatives, etc... which is to say, fear and excitement. Other emotional >responses can be argued similarly. > I agree that any thinking entity would have a negative reaction to a threat, involving some attempt to rapidly find alternatives. I just don't see this as being "fear" and "excitement". Let me explain why with an analogy: Why does a person take aspirin? I don't believe that the following goes on in his head -- "I say, It appears that those neurons over there are firing excessively. Perhaps I should interrupt their overly enthusiastic behavior..". I claim it is more like: "Owww... that really *hurts*. Gimme some aspirin... NOW!" Although the physical effect of the aspirin might be to cut off some signal in the nervous system, this has very little to do with a person's immediate motivation for taking it. I claim that the signal and the pain are two entirely different sorts of beasts. >"really feeling emotion." I'll be glad to answer this point, if you can first >prove to me that *you* "really feel" pain, sorrow, or pleasure, and don't just >react mechanically...) I've heard people (usually behaviorists) make this point but I'm never sure if they're serious (I didn't see a smiley :-). An attempt to answer the riddle of subjective experience by denying its existence seems somewhat pointless. BTW: In a torture situation, I don't think I would have a hard time convincing *anyone* that they can "really feel" pain. Would you agree that torture is wrong? Why? :-) > >/ I would rather not have a >/ machine that I would be afraid to turn off for fear of harming >/ *someone*. > >If you don't want a "someone", you'd better stay out of AI research... :-) > I am definitely *not* an opponent of AI. I think it is very likely that we will be able to create systems that are *operationally* indistinguishable from humans doing the same tasks. I think this will be a great thing. I do claim, however, that there is still likely to be a difference between an intelligent machine (here referring to a machine that models intelligent behavior in a functionalist sense, not by physically copying the brain) and a human (or other animal). >/ It does seem that our experience is rooted in some kind of >/ electro-chemical phenomenon, but I think it is an incredible leap of >/ faith to assume that logic circuits are all that is required :-). > >It's a point of faith, certainly, since we don't have more than one example. But >I don't think it's unjustified, since nothing more than logic circuits has ever >been observed. ("Logic circuits" is a bit of a misnomer, since neurons don't act >like single standard logic gates. However, it looks possible to simulate them >with electronics.) > >--Z As I mentioned earlier, I believe the standard functionalist approach to AI will bear fruit -- In fact, I think we will be able to generate systems to perform any tasks we can think of... even simulate a human! It seems unlikely that the same approach will generate artificial *beings* with subjective experience, but this is just fine with me. ;-) --Jerry Jackson
gibsong@gtephx.UUCP (Greggo) (12/01/88)
There are lots of good comments on this subject, but it's starting to degrade a bit into nit-picking on fine points of definitions, without attending to the main subject at hand. As to my views, I agree that learning and intelligence are related. However, much of the discussion has focused on technical definitions of intelligence. Don't emotions enter into intelligence at all, or do they just "get in the way"? One of the prime foundations for intelligence would surely be "an awareness of self". Most of the comments about considering whether the computer or the programmer understands assume a central point of control for intelligence. Are we intelligent first because we realize that we exist as an independent mind? How does this then apply to AI? Also, it is the _ability_ to learn, interpret environment, build experience, etc. that forms the foundation for intelligence, not the actual use. This explains how just because someone doesn't hear or care about what you're saying doesn't mean they're not intelligent. Again, this brings in attitudes and emotions, which at least influence our ability to exercise intelligence, if not directly a part of intelligence. In summary, some main ingredients of intelligence (one man's opinion): - awareness of self - ability to learn - emotions (curiosity, drive, satisfaction)? Anyway, I find this whole conversation fascinating. Please forgive the rambling nature of this posting. - greggo Disclaimer: Me, not GTE!
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (12/01/88)
In article <1983@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: >In article <1816@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: > >>Your not so subtle needling of Americans would seem less >>hypocritical if it originated in a country (and continent) > >As for cultural styles, I was only drawing attention to the >differences. I was not trying to create division or claim supriority. >As for approaches to literacy, well there are differences here, and it >looks like at least one American takes gentle jokey references >to this the wrong way. Still, this is an international net, and we >can't all tailor our style of writing to suit one culture. > Hmmm, well, ok. It looked to me like you were being pretty snide, but I'll take your word for it that you weren't. I certainly am quite familiar with British and European styles of humor, having spent a good deal of time there both as student and visiting professor. Anyhow I think it best we look more to the problems closer to home (of which there are plenty on both sides of the Atlantic) and leave the foreigners to generate their own social criticism from now on. >>Is this some supernatural notion, or is >>it the result of evolution, or culture, or some other natural process? >Don't know, do you? There is a whole range of experience which does >not seem to hav a mechanical basis. Which behaviour is AI trying to >cover (and do say 'intelligent' behaviour, since this means nothing here)? No, I don't know. I was the one asking the questions. From your strong statements that "humans are not machines" it appeared that you (at least thought you) had some answers. If by "not seem to have a mechanical basis" you mean that we can not duplicate the behavior (yet) or understand it mathematically, then fine, I agree. But prior to Newton, the same could be said for the motions of the planets. At least back to the time of Helmholtz, people began to realize that the body was a machine. That coupled with the idea that the brain is a physical system and is the locus of the mind and behavior, seems to me to indicate that there is a very significant probability that what we observe as our very complex behavior may be that of machines. This does not prove that this is so. Even if we could duplicate our behavior with machines of our own creation, one could never disprove absolutely that there wasn't some other force involved (perhaps God would send souls to occupy the machine bodies of the robots).
bwk@mitre-bedford.ARPA (Barry W. Kort) (12/01/88)
In article <281@esosun.UUCP> jackson@freyja.css.gov (Jerry Jackson) writes: > ... why does everyone >assume that an intelligent machine would be a *feeling* being? An ambulatory robot would be well advised to have a sensory alarm system to report mechanical stress in its limbs. Otherwise it is liable to damage itself while navigating through a hazardous environment. An intelligent machine that seeks to explore and learn about the world in which it is embedded would be well advised to have an emotional system which monitors its success or failure in knowledge acquisition. Successful lines of investigation would thereby be encouraged, while fruitless efforts would be abandoned in favor of a fresh tack. By monitoring its emotions, a learning system would also know whether to report its progress or ask for assistance when its mentor inquires, "How are you doing, today?" --Barry Kort
bwk@mitre-bedford.ARPA (Barry W. Kort) (12/01/88)
In article <1829@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes in response to Sarge Gerbode: >Then I presume that patients with various brain lesions, such as >bilateral lesions of the hippocampus are to be considered non-intelligent? Intelligence is not a binary trait. Last night on the PBS series, "The Mind", we saw how a stroke affected the mental life of a promising young attorney. The loss of function in his prefrontal lobes impaired his ability to conceive and plan a course of action and to solve problems. He is now a truck driver. After his stroke, it took a long time for the therapists to identify which faculties of intellect were lost. It is not yet clear whether the lost faculties can be reaquired. --Barry Kort
paulg@iisat.UUCP (Paul Gauthier) (12/02/88)
In article <281@esosun.UUCP>, jackson@freyja.css.gov (Jerry Jackson) writes: > In article <44150@yale-celray.yale.UUCP>, engelson@cs (Sean Philip Engelson) writes: > >all. They will have different needs, different feelings, different > > I can accept the needs, goals, and plans... but why does everyone > assume that an intelligent machine would be a *feeling* being? I see > no reason to assume that an IM would be capable of experiencing > anything at all. This doesn't imply that it wouldn't be intelligent. Actually, I think it does. Feelings are simply products of intelligence. Once any form of intelligence reaches the complexity of the human mind it will undoubtably experience 'feelings.' Feelings are simply manifestations of the minds goals and needs. You feel 'sad' when you don't attain a goal, this is simply a negative feedback response to prd you into trying harder. It might not work in all cases, but it helps. The word feeling is very broad. Feelings of fear are manifestations of your minds attempts to deal with the unkown or threats. What you experience as fear is the workings of your mind trying to come to a decision in a tough situation. This whole topic is very hard to discuss and I'm sure I've bungled it quite nicely, but I hope I have put accross something resembling my true opinion on this. All this things people refer to as feelings, things which many consider for humans-alone, are results of inconsistancies in our knowledge-bases and signs of our intelligense working. A feeling is an educated guess that our mind makes based on what it can puzzle out from known facts. As you can see, the word 'feeling' doesn't do well to classify all the myriad of types of feeling there are so it is hard to discuss... > For instance: some machines are already capable of distinguishing blue > light from red. This doesn't mean that they have anything like our > *experience* of blue. (Or pain, or sorrow, or pleasure... etc.) All your *experience* of blue is is your brain searching its memory to figure out what 'blue' is. Undoubtably it flashes through memories connected to 'blue' which trigger the *experience* of blue. When machines have large enough inter-connected knowledge-bases they too will come accross past experiences which relate to blue and *experience* the color. > Personally, I think this is a good thing. I would rather not have a > machine that I would be afraid to turn off for fear of harming > *someone*. It does seem that our experience is rooted in some kind of > electro-chemical phenomenon, but I think it is an incredible leap of > faith to assume that logic circuits are all that is required :-). Personally, I find the idea exciting. I'm patiently waiting for the first machine sentience to emerge. I feel it is possible, and it is only a matter of time. After all, humans are only carbon-based machines. > > --Jerry Jackson -- |=============================================================================| | Paul Gauthier: {uunet, utai, watmath}!dalcs!iisat!{paulg | brains!paulg} | | Cerebral Cortex BBS: (902)462-7245 300/1200 24h/7d N81 | |==============================================================================
hawley@icot32.icot.junet (David John Hawley) (12/02/88)
In article <2732@uhccux.uhcc.hawaii.edu> lee@uhccux.uhcc.hawaii.edu (Greg Lee) writes: >From article <562@metapsy.UUCP>, by sarge@metapsy.UUCP (Sarge Gerbode): >" ... >" Do machines have the same subjective experience that we do when we ... >" and the input data, Occam's Razor demands that we not attribute >" subjectivity to them. > >A more proper application of Occam's Razor would be that it prevents >us from assuming a difference between humans and machines in this >regard without necessity. What does explaining behavior have to ... What are the criteria by which I may judge the suitability of an application of Occam's Razor? I know the folktale is basically the KISS principle, and I have heard that the actual criterion of simplicity is the number of 'blat' that need to be postulated (where a blat can be a thing, entity, ?property?, ...). Is this correct? This has something to do with theory formation, as per for example David Poole's Theorist default-reasoning system. Does anyone have any pointers to literature on theory preference, relative strength of arguments, preferably in an how-could-we-build-it vein? Yoroshiku (AdvTHANKSance) David Hawley
ok@quintus.uucp (Richard A. O'Keefe) (12/02/88)
In article <3ffb7cfc.14c3d@gtephx.UUCP> gibsong@gtephx.UUCP (Greggo) writes: >Don't emotions enter into intelligence at all, or >do they just "get in the way"? Emotions have often been discussed in the AI literature. See, for example, Aaron Sloman's "You don't need a soft skin to have a warm hear." Emotions have a large cognitive component; they aren't just physiological. (C.S.Lewis in his essay "Transposition" pointed out that Samuel Pepys reported the same phsyical sensations when seasick, when in love with his wife, and on hearing some wind music, and in the latter case promptly decided to practice the instrument.) Considering the range of human temperaments, the experience and expression of emotion probably isn't necessary for "intelligence". I wonder, though. I have seen programs which nauseated me, and they were bad programs, and I have seen programs which brought tears of pleasure to my eyes, and they were good programs. If emotions can be aroused by such "mathematical" things as programs, and aroused *appropriately*, perhaps they are more important than I think. Such emotions certainly motivate me to write better programs. >One of the prime foundations for >intelligence would surely be "an awareness of self". "Foundation" in what sense? Let's be science fictional for a moment, and imagine a sessile species, which every spring buds off a mobile ramet. The mobile ramet sends sense data to the sessile ramet, and the sessile ramet sends commands to the mobile one. The sessile ramets live in a colony, and the mobile ones gather food and bring it back, and otherwise tend the colony. Every winter the mobile ramets die and the sessile ones hibernate. The mobile ramets are "cheap" to make because they have just enough brain to maintain their bodies and communicate with the sessile ones, which means that they can be quite a bit smaller than a human being and still function intelligently, because the brain is back in the sessile ramet. Is it necessary for the sessile ramet to know which of the ones in the colony is itself? No, provided all the sessiles are maintained, it doesn't much matter. (It helps if physiological states of the sessiles such as hunger and illness are obvious from the outside, wilting leaves or something like that.) These creatures would presumably be aware of themselves *as*mobiles*. I was about to write that an intelligent entity would need access to its own plans in order to critise them before carrying them out, but even that may not be so. Imagine a humanoid robot which is *not* aware of its own mental processes, but where information about those processes is visible on a debugging panel at the back. Two such robots could check each other without being able to check themselves. "An awareness of self" might be important to an intelligent organism, but it might be a *consequence* of intelligence rather than a *precondition* for it. It is usually claimed that human babies have to learn to distinguish self from non-self. (How anyone can _know_ this I've often wondered.)
ok@quintus.uucp (Richard A. O'Keefe) (12/03/88)
In article <177@iisat.UUCP> paulg@iisat.UUCP (Paul Gauthier) writes: >Feelings are >simply manifestations of the mind's goals and needs. You feel 'sad' when >you don't attain a goal, this is simply a negative feedback response to >prd you into trying harder. It might not work in all cases, but it helps. If I heard that you had died unexpectedly, I would feel sad. But the preservation of your life would not previously have been one of my goals. (I would, for example, be unable to distinguish you from Gilbert Cockton.) There's an interesting question here about human psychology: which emotions are innate, and which emotions are culture-dependent. Lakoff's book includes a list of apparently culture-independent emotional states (unfortunately I left W,F&DT at home, so I can't quote it), and I was surprised at how short it was. One anthropological commonplace is the distinction between "guilt" cultures and "shame" cultures. Most humans presumably have the capacity to learn these emotions, but apparently we do not all experience the same emotions in the same circumstances. (Which emotions a culture has basic _terms_ for is another question.)
raino@td2cad.intel.com (Rodger Raino) (12/03/88)
Hi I'm an outsider to this group, but our newsfeed seems blocked so I stumbled into this lively discussion. You guys are having fun arguing it looks like. I thought I'd let you know how it's *really* gonna work out. Now most of you are going to read this and go, "boy is that simpleminded/dumb/naive" but just stick this in the back of your head and remember it in about fifty years. You'll see. Shortly lots of "smart" traffic lights are going to start talking to each other upstream and downstream to make a smoother flow of traffic. After awhile this will set up a bunch of resonance patterns in the flow of traffic. This in turn will become coupled to a host of "other" information passing networks (ei. drivers = capacitors, you're now always turing (triple pun intended) on Entertainment Tonight at exactly the same time). What you end up with is a vast complex and fuzzy network that will spontaneously become intelligent. Of course this "new" entinity mignt not be able to feed the ducks, but then the duck feeder can't decide to power down the planet either. cheers rodger -- -_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_-_ i know i'm a bad speller, don't waste FlameWidth pointing out old news intel agrees with this, but not necessarly anything above the line. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
chrispi@microsoft.UUCP (Chris Pirih) (12/03/88)
In article <42328@linus.UUCP> bwk@mbunix (Kort) writes: >Andrew C. Plotkin writes: >>I maintain that a human can be simulated by a Turing machine. Comments? > ... I use a quantum amplifier in my coin flipper. >Correct me if I'm wrong. But a Turing Machine is obliged to follow >a deterministic program. Hence a Turing machine cannot simulate my >dice-tossing method. Nothing prevents the Turing machine from flipping a coin and acting [deterministically] on the [random[?]] result. (What exactly are we trying to simulate here, Barry? Is the coin, with its quantum randomness, a part of the human who consults it?) Besides, is it necessary that a simulated coin-flip be "truly" random, or just effectively unpredictable (i.e., that the Turing android eschew foreknowledge of its pseudo-random pseudo-coin-flip)? The latter seems sufficient to me. --- chris (Besides, I never toss dice to make a decision; your Turing machine should have no problem simulating me...)
maddoxt@novavax.UUCP (Thomas Maddox) (12/04/88)
In article <800@quintus.UUCP> ok@quintus.UUCP (Richard A. O'Keefe) writes: >There's an interesting question here about human psychology: which >emotions are innate, and which emotions are culture-dependent. Lakoff's >book includes a list of apparently culture-independent emotional states >(unfortunately I left W,F&DT at home, so I can't quote it), and I was >surprised at how short it was. From Lakoff, _Women, Fire and Dangerous Things_, p. 38: In a major crosscultural study of facial gestures expressing emotion, Ekman and his associates discovered that there were basic emotions that seem to correlate universally with facial gestures: happiness, sadness, anger, fear, surprise, and interest. Of all the subtle emotions that people feel and have words and concepts for around the world, only these have consistent correlates in facial expressions across cultures. end-quote I agree that Lakoff's book is extremely interesting with regard to key problems in AI, particularly in its replacement of what he calls the "classical theory that categories are defined in terms of common properties of their members" with a new view ("experimental realism" or "experientialism"). In his "Preface," Lakoff says, The issue is this: Do meaningful thought and reason concern merely the manipulation of abstract symbols and their correspondence to an ojbective reality, independent of any embodiment (except, perhaps, for limitations imposed by the organism)? Or do meaningful thought and reason essentially concern the nature of the organism doing the thinking--including the nature of its body, its interactions in its environment, its social character, and so on? end-quote Like Lakoff, I'm convinced that the second set of answers points in the correct direction. As a science fiction writer who has tried to present an artificial intelligence realistically, I saw from the start that the *embodied* categories Lakoff speaks of had to be presupposed in order to present a being I could consider intelligent. (By the way, I hope readers see there is a difference between Lakoff's view, which poses interesting questions for AI research, and the views of eminent anti-AI theorists such as Dreyfus and Weizenbaum [and vocal net anti-AI types such as Cockton]. Lakoff says (p. 338): I should point out that the studies discussed in this volume do not in any way contradict studies in artificial intelligence . . . in general. . . . We shall discuss only computational approaches to the study of mind. Even there, our results by no means contradict all such approaches. For example, they do not contradict what have come to be called "connectionist" theories, in which the role of the body in cognition fits naturally. end-quote) Lakoff's work is especially interesting when set next to a recent book by Terry Winograd and Fernando Flores, _Understanding Computers and Cognition_. They likewise reject the tradition which sees reason as "the systematic manipulation of representations." However, they use a philosophical tradition very different from that employed in usual AI studies: to wit, the Continental tradition of hermeneutics and phenomenology that includes Heidegger and Gadamer. They also include "speech act" theory, from Austin and Searle and, in biology, Maturana's work. What these books (along with some essays of Daniel Dennett's) represent to me is an attempt at coming to terms conceptually with the high-level problems posed by AI. The doctrinaire anti-AI group continue to snipe from the sidelines, with arguments that say (1) it can't be done, and (2) even if it could, it shouldn't; the workers who are trying to create artificial intelligence (i.e., the makers of the hardware and software) quite often are submersed entirely in their particular problems and speak almost exclusively in the technicalities appropriate to those problems. Thus, the intelligent and approachable work being done by Lakoff et alia serves us all: this is one of the characteristic problems of our time and one of our civilization's greatest wagers, and those of us who are trying to understand it (rather than deride or implement it) need a coherent universe of discourse in which understanding might take place.
ap1i+@andrew.cmu.edu (Andrew C. Plotkin) (12/04/88)
/>Aren't feeling and emotion automatic byproducts of thought? I find it hard to
/>imagine an entity capable of thought which would *not* have (for example) some
/>negative reaction to a threat, involving some attempt to rapidly find
/>alternatives, etc... which is to say, fear and excitement. Other emotional
/>responses can be argued similarly.
/>
/
/ I agree that any thinking entity would have a negative reaction to a threat,
/ involving some attempt to rapidly find alternatives. I just don't see this
/ as being "fear" and "excitement". Let me explain why with an analogy:
/
/ Why does a person take aspirin? I don't believe that the following
/ goes on in his head -- "I say, It appears that those neurons over
/ there are firing excessively. Perhaps I should interrupt their overly
/ enthusiastic behavior..". I claim it is more like: "Owww... that really
/ *hurts*. Gimme some aspirin... NOW!" Although the physical effect of
/ the aspirin might be to cut off some signal in the nervous system, this
/ has very little to do with a person's immediate motivation for taking it.
/ I claim that the signal and the pain are two entirely different sorts of
/ beasts.
Even today we have computer programs that have no "idea" (no access to) what
goes on in their lower levels (the machine language.) A Lisp program manipulates
lists without ever referring to the RAM addresses they're stored at. This seems
to me to be an exact equivalent (much simpler of course) to the way we don't
worry about what neurons are firing. If an AI is written in a high-level
language, I would expect that it would have no idea of what routines are running
it. Similarly, if an AI is developed by making a big neural net and kicking it,
it would not know what sort of patterns are running around its circuits. It
would just react by saying things like "OW!"
/ I think we will be able to generate
/ systems to perform any tasks we can think of... even simulate a human!
/ It seems unlikely that the same approach will generate artificial
/ *beings* with subjective experience, but this is just fine with me. ;-)
You mentioned torture -- if you had a computer console in front of you with a
button marked "pain" (any human simulator had better have some sort of sensory
input), would you consider it okay to push it? How about if a screen (or
speaker) was printing the output produced by the program as it did a nice
simulation of a human begging you to stop? If you first spent an hour or so
discussing your favorite music or movies? (Excuse me; I mean "using the
simulation to see how the human being simulated would respond to your opinions
on music or movies.")
Yes, I know, that paragraph is intended to play on your sympathy. But
consider it seriously. You go into a room, spend a while talking to and getting
answers from a computer console. Being a produced by a good human simulation,
the conversation is typical of any you'd have with a random stranger. Would you
then feel morally justified pushing that button, saying "it's only a simulation"?
/>"really feeling emotion." I'll be glad to answer this point, if you can first
/>prove to me that *you* "really feel" pain, sorrow, or pleasure, and don't just
/>react mechanically...)
/
/ I've heard people (usually behaviorists) make this point but I'm never sure
/ if they're serious (I didn't see a smiley :-). An attempt to answer the
/ riddle of subjective experience by denying its existence seems somewhat
/ pointless.
I'm not -denying- subjective experience. I'm saying that, -whatever subjective
experience is-, if system X and system Y both act the same, it's silly to say X
has subjective experience and Y doesn't. Especially when the only difference is
that X grew by itself and Y was built by X.
This subject has been covered by more impressive people than me; dig up _The
Mind's I_ by Hofstadter and Dennet, which has many essays (and fiction) by lots
of people and commentary by H & D, all on AI and minds and brains and stuff. Fun
to read, too.
--Zsarge@metapsy.UUCP (Sarge Gerbode) (12/04/88)
In article <2732@uhccux.uhcc.hawaii.edu> lee@uhccux.uhcc.hawaii.edu (Greg Lee) writes: >From article <562@metapsy.UUCP>, by sarge@metapsy.UUCP (Sarge Gerbode): >" Since [machines'] behavior is completely >" explainable in terms of the hardware design, the software program, >" and the input data, Occam's Razor demands that we not attribute >" subjectivity to them. > >A more proper application of Occam's Razor would be that it prevents >us from assuming a difference between humans and machines in this >regard without necessity. What does explaining behavior have to >do with it? If I could explain your behavior, would this have the >consequence that you cease to have subjective experience? Of course >not. (If *you* could explain your behavior, perhaps the case could >be made ...) I don't need a mechanistic explanation of my own behavior (much of it, at least), because I am directly aware of causing it by intention. Furthermore, the most major observable difference between myself and a machine is that the latter is explainable in mechanistic terms, whereas I am not. On the other hand, if I could explain *your* behavior entirely on mechanistic grounds, then I think I would have grounds (Occam's Razor) for not attributing subjectivity to you. However, I don't think I can do that, and so I don't think you are a machine. It is because others are observably *not* machines, not explainable in mechanistic terms, that we attribute subjectivity (and humanity) to them, in order to explain their behavior. People don't like to be manipulated, programed, treated like machines, and part of the reason why is, I believe, that they have an immediate awareness of themselves as not being mechanistically determined, and that sort of treatment observably embodies a lie. -- -------------------- Sarge Gerbode -- UUCP: pyramid!thirdi!metapsy!sarge Institute for Research in Metapsychology 950 Guinda St. Palo Alto, CA 94301
ap1i+@andrew.cmu.edu (Andrew C. Plotkin) (12/05/88)
/In article <42328@linus.UUCP> bwk@mbunix (Kort) writes: />Andrew C. Plotkin writes: /> >I maintain that a human can be simulated by a Turing machine. Comments? /> /> ... I use a quantum amplifier in my coin flipper. />Correct me if I'm wrong. But a Turing Machine is obliged to follow />a deterministic program. Hence a Turing machine cannot simulate my />dice-tossing method. chrispi@microsoft.UUCP (Chris Pirih) replies... /Nothing prevents the Turing machine from flipping a coin and acting /[deterministically] on the [random[?]] result. (What exactly are we /trying to simulate here, Barry? Is the coin, with its quantum /randomness, a part of the human who consults it?) Besides, is it /necessary that a simulated coin-flip be "truly" random, or just /effectively unpredictable (i.e., that the Turing android eschew /foreknowledge of its pseudo-random pseudo-coin-flip)? I agree with Chris here. When I flip a coin in my head, I have serious doubts that the results are truly random (based on an amplified quantum randomness.) It might just as well be a complex pseudo-random generator. It would work just as well for any practical purpose, such as the balanced ethical dilemma (sp?) you mentioned. (Practical purpose meaning anything where you need a single random bit. If you try to spit out a long string of random bits, the non-randomness of the process becomes painfully clear -- there is a strong correlation between each bit and the several-bit sequence that precedes it. This can be checked with a simple computer program (on each bit, using the previous four or five bits to predict the next). I've found it a 60% to 70% accurate predictor; no doubt a more complex program could do better.) --Z
bwk@mitre-bedford.ARPA (Barry W. Kort) (12/05/88)
In article <1069@microsoft.UUCP> chrispi@microsoft.UUCP (Chris Pirih) joins the discussion on quantum randomness vs. Turing-computable randomness: > ... Besides, is it > necessary that a simulated coin-flip be "truly" random, or just > effectively unpredictable (i.e., that the Turing android eschew > foreknowledge of its pseudo-random pseudo-coin-flip)? The latter > seems sufficient to me. It is enough that the random number generator be unpredictable (by *any* predictor, including one who would cheat and examine my random number generation process). The only way I know to do this is to use a quantum amplifier, so that no one can anticipate the outcome (myself included). A Turing machine can compute a pseudo-random sequence, but it cannot implement a random number generator based on a quantum amplifier. (Such a device would cease to be a Turing Machine.) --Barry Kort
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (12/05/88)
In article <563@metapsy.UUCP> sarge@metapsy.UUCP (Sarge Gerbode) writes: >I don't need a mechanistic explanation of my own behavior (much of >it, at least), because I am directly aware of causing it by >intention. Furthermore, the most major observable difference between >myself and a machine is that the latter is explainable in mechanistic >terms, whereas I am not. Neither of these two propositions can be demonstrated reliably. The behaviorists have shown that behavior which subjectively seems to us to be caused by intention can be determined (even hypnotists can demonstrate this), therefore your impressions are unreliable. In addition, a complex enough neural network can demonstrate behavior the cause of which is not immediately apparent. Obviously no network has been invented as complex as the human brain, and until one is we won't be able to answer the question experimentally. Those bothered by possible loss of free will should recall that in a system complex enough, there is room for the possibility of indeterminacy, be it a biological system or whatnot. I will ask Serge the same questions I asked Gilbert: if humans are not a machine, what elements are added to the body (which seems to be a physical machine as far as we can tell) which make it otherwise? Are these material or immaterial? Is there some aspect of human beings which does not obey the laws of nature?
hajek@gargoyle.uchicago.edu.UUCP (Greg Hajek) (12/06/88)
In article <1841@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >Those bothered >by possible loss of free will should recall that in a system complex >enough, there is room for the possibility of indeterminacy, be >it a biological system or whatnot. Well, it's not immediately apparent that indeterminacy is a function of complexity, in any sense. The two-slit experiment is extremely simple, analyzed within the context of quantum mechanics, but that doesn't resolve the question of point-wave duality. Similarly, no PDP network will exhibit behavior that defies a deterministic explanation when run on a computer; indeed, just dump every step of processing, and you have a low-level explanation right there (of course, as complexity increases, you increase the possibility that, say, a cosmic-ray particle will come screaming through your computer, but even such an event as this is not "indeterminate"). >I will ask Serge the same questions I asked Gilbert: if humans are >not a machine, what elements are added to the body (which seems to >be a physical machine as far as we can tell) which make it otherwise? >Are these material or immaterial? Is there some aspect of human >beings which does not obey the laws of nature? I wasn't asked, but while I'm shooting my mouth off . . . if humans are not machines, of course there is no material addition to the body, since that would just comprise a different machine. Nor is there any assumption that humans do not obey the laws of nature, but rather that our perspective on the laws of nature as being equivalent to the "laws" of physics is erroneous. This is required from a dualist point of view, for instance: if a non-physical event can govern physical behavior, conservation of energy goes right out the window, and not just such that (delta E)*(delta t) <= h. But assuming such a dualist stance leaves some ugly questions: why do the empirical observations of physicists generalize so well to encompass new situations (that is, why does physics work)? If a nonphysical theory of the world is going to be adopted, it's really not a good enough reason to do so just so that the task of creating an intelligent machine is impossible, sans any other motivation. I would certainly expect God to pick on the economists, too, at least. ---------- Greg Hajek {....!ihnp4!sphinx!gargoyle!hajek} "I don't know what the big deal is, I don't feel anything yeeeEEEEAAAAAA...."
bwk@mitre-bedford.ARPA (Barry W. Kort) (12/06/88)
In article <563@metapsy.UUCP> sarge@metapsy.UUCP (Sarge Gerbode) writes: > I don't need a mechanistic explanation of my own behavior (much of > it, at least), because I am directly aware of causing it by > intention. I agree with Sarge that, as a human being, I frequently engage in goal-seeking behavior. That is, I have intentions. I also engage (from time to time) in goal-choosing behavior. But unlike my goal-seeking behavior, my goal-choosing behavior seems much more unintentional. Sometimes goals are thrust upon me by circumstances or cultural expectations. Sometimes goals surface as part of a natural progression of learning (as in research). In any event, I find it hard to predict what goals I will adopt after I complete my current agenda. (But I also suspect that a more sagacious soul than I would have less trouble ancticipating my future goals.) --Barry Kort
bwk@mitre-bedford.ARPA (Barry W. Kort) (12/06/88)
In article <wXaC5Ky00V4U05hEUM@andrew.cmu.edu> ap1i+@andrew.cmu.edu (Andrew C. Plotkin) writes: > ... if you had a computer console in front of you with a button > marked "pain" (any human simulator had better have some sort of sensory > input), would you consider it okay to push it? Yes, if I was testing the computer's pain circuits. When a computer is in pain (i.e. a circuit board is burning out, or a cable is being cut), I want to be sure that it can sense its distress and accurately report its state of well-being. Similarly, if I put the machine in emotional pain (by giving it a program that runs forever and does no useful work), I hope the machine can diagnose the problem and gracefully aprise me of my error. Getting an incomprehensible core dump is like having a baby throw up because something it ate was indigestible. (I find core dumps indigestible.) Barry Kort
jeff@censor.UUCP (Jeff Hunter) (12/06/88)
In article <177@iisat.UUCP>, paulg@iisat.UUCP (Paul Gauthier) writes: > In article <281@esosun.UUCP>, jackson@freyja.css.gov (Jerry Jackson) writes: > > I can accept the needs, goals, and plans... but why does everyone > > assume that an intelligent machine would be a *feeling* being? I see > > no reason to assume that an IM would be capable of experiencing > > anything at all. This doesn't imply that it wouldn't be intelligent. > Actually, I think it does. Feelings are simply products of > intelligence. Once any form of intelligence reaches the complexity of > the human mind it will undoubtably experience 'feelings.' Feelings are > simply manifestations of the minds goals and needs. You feel 'sad' when I disagree. Emotions form a relatively simple reasoning system. (Lest you get the wrong impression from the start let me hasten to add that I enjoy my emotions [most of the time at least]. I'm just saying that they're not all that bright.) For example : I like you. I share with you. You like me. You share back. There's a viable society without needing deep thought on the economics of co-operation vs competition, or long computer modelling runs, etc... "Like", "trust", "distrust", and "hate" form such useful models of behaviour that just about any mammal or bird use them to reason about relationships. I assume that any relatively dumb intelligences that need to operate in some social environment would need some similar shortcuts to reason with. Smarter intelligences "evolved" from the dumb ones would probably retain the emotions just from design economy. Emotional reasoning can often outperform logical reasoning (watch any episode of Star Trek :-). Lots of people have stopped smoking because of guilt rather than reasoned argument. However emotions (especially strong ones) can make people do really dumb things too. Blind love and blinding hatred are cliches. If I was dealing everyday with an artificial intelligence then I'd prefer it to have human-like emotions (or at least dog-like). I'd make an emotional attachment and I'd be sort-of disappointed if it declared me surplus protein and fed me to the grinder :-). However an intelligence that didn't have to interact with others wouldn't need to be run by emotions. A robot asteroid miner, for example, could be set the task of converting a planetoid into can openers and paper weights. It wouldn't have to have a favourite ore truck, or be pleased with the day's output, or panic and freeze if a major mechanical failure happens. It wouldn't even have to feel disappointed or noble as it melts itself down to make the last crate of paper weights. Conversely I could see an emotional version of the same machine that could probably do just as good a job. (The emotions would have to be adjusted from human norms though.) In summary I think that intelligence doesn't require emotions, but near-term "real" artificial intelligences will need them to interact with humans, and the emotions will probably hang around unless they are degined out for a purpose. -- ___ __ __ {utzoo,lsuc}!censor!jeff (416-595-2705) / / /) / ) -- my opinions -- -/ _ -/- /- The first cup of coffee recapitulates phylogeny... (__/ (/_/ _/_ Barry Workman
pepke@loligo.fsu.edu (Eric Pepke) (12/06/88)
A Turing machine cannot just consult a truly random coin flipper. If it could, it wouldn't be a Turing machine. However, there is a much simpler objection to the argument. A random number generator can only be consulted a finite number of times in a lifetime. For every finite sequence of such random numbers, you can produce a partial Turing machine specification which produces that sequence. So, there's no problem. -EMP
ok@quintus.uucp (Richard A. O'Keefe) (12/06/88)
In article <1841@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >Neither of these two propositions can be demonstrated reliably. >The behaviorists have shown that behavior which subjectively seems >to us to be caused by intention can be determined (even hypnotists >can demonstrate this), Er, how do hypnotists demonstrate that? Perhaps I've read too many issues of the Skeptical Inquirer and not enough of the National Enquirer, but my understanding was that hypnotism is these days regarded as a form of voluntary fantasy. (We'll just have to put up with "voluntary" until Skinner sends me the Official Phrasebook.) As for the first part of this, there is a philosophical tradition called "compatibilism", which holds that "it was caused by intention" and "it was determined" are not contradictory.
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (12/06/88)
In article <286@gargoyle.uchicago.edu> hajek@gargoyle.uchicago.edu.UUCP (Greg Hajek) writes: > >Well, it's not immediately apparent that indeterminacy is a function of >complexity, in any sense. Simple systems of macroscopic dimensions are clearly deterministic, would you agree? Thus, any hope for indeterminacy lies in the complexity of a system being such that the non-analyticity is guaranteed. For example, when you stand at the base of a waterfall, you will from time to time be splashed by jets of water. But you can not mechanistically compute when you will be splashed because of the complexity of the system. >Similarly, no PDP network will exhibit behavior >that defies a deterministic explanation when run on a computer; indeed, just >dump every step of processing, and you have a low-level explanation right there Ah, but the low level explanation may not make any sense of the behavior, but only describes it. Making sense of it requires interpretation. Take simple backprop programs, for example. The experimenter knows what the input and output units are to be, but does not determine the final successful configuration of the hidden units. Often, their final state is a surprise, but still makes sense after interpretation. > >>I will ask Serge the same questions I asked Gilbert: if humans are >>not a machine, what elements are added to the body (which seems to >>be a physical machine as far as we can tell) which make it otherwise? >>Are these material or immaterial? Is there some aspect of human >>beings which does not obey the laws of nature? > >I wasn't asked, but while I'm shooting my mouth off . . . if humans are not >machines, of course there is no material addition to the body, since that would >just comprise a different machine. Nor is there any assumption that humans do >not obey the laws of nature, but rather that our perspective on the laws of >nature as being equivalent to the "laws" of physics is erroneous. This is >required from a dualist point of view, for instance: if a non-physical event >can govern physical behavior, conservation of energy goes right out the window, >and not just such that (delta E)*(delta t) <= h. > This recapitulates Helmholz' reasoning when he decided that conservation of energy required humans to be machines. I have yet to see anyone make a satisfactory argument to the contrary. Obviously, if one brings religion or magic into the equation then it opens many possibilities, but so far no one in this discussion has cited either of those as their reasons for denying that humans are machines.
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (12/07/88)
In article <817@quintus.UUCP> ok@quintus.UUCP (Richard A. O'Keefe) writes: >In article <1841@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >>Neither of these two propositions can be demonstrated reliably. >>The behaviorists have shown that behavior which subjectively seems >>to us to be caused by intention can be determined (even hypnotists >>can demonstrate this), > >Er, how do hypnotists demonstrate that? People who were hypnotized usually report not that they were compelled to perform the suggested actions but that they "felt like it". In other words, the subjective impression was that the actions were voluntary, yet they do ridiculous things that are clearly determined by the suggestion. If you wish to claim that post-hypnotic suggestions are true free-will voluntary actions, then I can only argue with your definition.
ray@bcsaic.UUCP (Ray Allis) (12/07/88)
ok@quintus.uucp (Richard A. O'Keefe) says: >"An awareness of self" might be important to an intelligent organism, >but it might be a *consequence* of intelligence rather than a >*precondition* for it. "Self-awareness" is not required for intelligent behavior (my definition of intelligence, of course), but it IS necessary for the association of experiences which constitutes symbol definition. Note that symbol *manipulation*, as by digital computers, can proceed in the total absense of intelligence.
rjc@aipna.ed.ac.uk (Richard Caley) (12/07/88)
In article <563@metapsy.UUCP> sarge@metapsy.UUCP (Sarge Gerbode) writes: >I don't need a mechanistic explanation of my own behavior (much of >it, at least), because I am directly aware of causing it by >intention. It is precisly because of this I need an expalanation. I have a reasonably strong belief in causality and without an explanation I am reduced to talking about "intention" or something similar as a magical something to break the causal chain at me. >Furthermore, the most major observable difference between >myself and a machine is that the latter is explainable in mechanistic >terms, whereas I am not. On the other hand, if I could explain >*your* behavior entirely on mechanistic grounds, then I think I would >have grounds (Occam's Razor) for not attributing subjectivity to >you. Come on, I can not explain al the beaviour of the global weather system on mechanistic grounds - in fact, nobody can at this time. does this mean we have to go back to rain gods and thunder gods, atributing 'subjectivity' to the weather. I would chalange you to expalain the behaviour of the computer on which you read this on mechanistic grounds. I can't. In fact until the physicists have solved all tehir problems no one will be able to. By your own argument. . . . >It is because others are observably *not* machines, >not explainable in mechanistic terms, that we attribute subjectivity >(and humanity) to them, in order to explain their behavior. Defining subjectivity as humanity is rather a good way of wining the argument. However, this is usually called "cheating". I can prove that it is impossible to sit on a stool by defining the property of being a seat as being "chairness" and so since stools are not chairs . . . >People don't like to be manipulated, programed, treated like machines, >and part of the reason why is, I believe, that they have an immediate >awareness of themselves as not being mechanistically determined, and >that sort of treatment observably embodies a lie. This does not make them correct. People believe all sorts of things. I believe that pushing something twice as hard makes it go twice as fast. This is incorrect, but it gets me around the supermarket with my trolly so what the hell. -- rjc@uk.ac.ed.aipna AKA rjc%uk.ac.ed.aipna@nss.cs.ucl.ac.uk "We must retain the ability to strike deep into the heart of Edinburgh" - MoD
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (12/07/88)
From article <163@censor.UUCP>, by jeff@censor.UUCP (Jeff Hunter): " " I disagree. Emotions form a relatively simple reasoning system.... " There's a viable society without needing deep thought on the economics of " co-operation vs competition, or long computer modelling runs, etc... This is a parochially human view. Emotions seem simple to you, because you can experience them without effort or reflection, and because "lower" animals have them too. Why not reason that since the mechanisms that permit animals to form societies took much longer to evolve than those permitting human reason that the former must be more complex? I think that would make at least as much sense. Greg, lee@uhccux.uhcc.hawaii.edu
ap1i+@andrew.cmu.edu (Andrew C. Plotkin) (12/08/88)
/ > ... if you had a computer console in front of you with a button
/ > marked "pain" (any human simulator had better have some sort of sensory
/ > input), would you consider it okay to push it?
/ Yes, if I was testing the computer's pain circuits. When a computer
/ is in pain (i.e. a circuit board is burning out, or a cable is being
/ cut), I want to be sure that it can sense its distress and accurately
/ report its state of well-being.
Ah, but I'm not talking about a system that senses damage to the computer. I'm
talking about something that applies stimuli to the simulated pain inputs of the
simulated human.
You brought up "computers being able to simulate humans," and I'm using that
concept. To clarify it, let me describe it as a program running on a computer;
with input routines that feed data to the same thought-mechanisms that human
sensory nerves feed to in the human mind; with output routines that take data
from the appropriate thought-mechanisms and display it in suitable form. Given
any input, it will produce output as a typical human would. (Passing the Turing
test, therefore.)
(The "easiest" way to this is to create a trillion CPU's, each capable of
simulating one neuron, and hooking them together. Sensory input could then be
pushed into the "sensory neurons" directly. However, the exact mechanism is not
relevant here.)
Now, there's a big difference between damage to the computer and simulated
pain. One degrades the performance of the simulation; the other makes the
simulation yell "ouch!" (assuming it's a good simulation.)
Obvious example: if a brain surgeon is working on a conscious patient, the
patient feels no pain (assuming the cut scalp has been numbed.) The surgeon can
poke around, feed minute electrical currents in, and so forth; the patient will
see strange flashes, have odd bits of memory pop up, and so forth. If the
surgeon drops his scalpel in, the patient will stop thinking or suffer
functional loss, but no pain is involved, unless sensory centers are hit.
/ Similarly, if I put the machine in
/ emotional pain (by giving it a program that runs forever and does
/ no useful work), I hope the machine can diagnose the problem and
/ gracefully aprise me of my error.
Keep thinking human simulation. The machine would simulate reactions like "Damn,
this is boring." Or, more likely, "Why should I do this idiot work? Program it
into a computer!"
(Of course, if it was a simulation of a reasonably open-minded human, you
could easily convince it that it was really a computer. That its optical inputs
are coming from cameras would be a giveaway. But I doubt it was settle down and
execute C for the rest of its existence. Assume it was a simulation of you --
would you?)
--Zok@quintus.uucp (Richard A. O'Keefe) (12/08/88)
In article <1847@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >In article <817@quintus.UUCP> ok@quintus.UUCP (Richard A. O'Keefe) writes: >>In article <1841@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >>>The behaviorists have shown that behavior which subjectively seems >>>to us to be caused by intention can be determined (even hypnotists >>>can demonstrate this), >> >>Er, how do hypnotists demonstrate that? > >People who were hypnotized usually report not that they were >compelled to perform the suggested actions but that they "felt >like it". In other words, the subjective impression was that the >actions were voluntary, yet they do ridiculous things that are clearly >determined by the suggestion. If you wish to claim that post-hypnotic >suggestions are true free-will voluntary actions, then I can >only argue with your definition. I note that Banks didn't quote the bit where I pointed out that hypnosis is understood these days as a sort of voluntary fantasy: the subject does what s/he thinks a hypnotic subject ought to do. To say that the actions "are clearly _determined_ by the suggestion" begs the question. How would you show that an action performed in response to a post- hypnotic suggestion is not voluntary? (Anyone who wants to claim "I was hypnotised" as a defence in court had better be prepared for a nasty surprise.) The thing is, being-a-hypnotic-subject is a social context in which it is acceptable, even *expected*, for the subject to "do ridiculous things". Try instead a hypnotic experiment where the subjects are told in advance that if for every commmand they obey they will be fined $100, or try a stage hypnotist's act with an audience of confederates who boo whenever the subject does something silly. Instead of arguing with my definition of "voluntary", it might be better to read up on hypnotism in the scientific literature.
bwk@mitre-bedford.ARPA (Barry W. Kort) (12/08/88)
In article <180@loligo.fsu.edu> pepke@loligo.UUCP (Eric Pepke) writes: > A random number generator can only be consulted a finite number of > times in a lifetime. For every finite sequence of such random numbers, > you can produce a partial Turing machine specification which produces > that sequence. So, there's no problem. Just one problem, Eric. You have to build your Turing Machine emulator before I have finished living my life. The information you need to construct it is not available just yet. --Barry Kort
bwk@mitre-bedford.ARPA (Barry W. Kort) (12/08/88)
In article <1841@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: > In addition, a complex enough neural network can demonstrate behavior > the cause of which is not immediately apparent. Obviously no network > has been invented as complex as the human brain, and until one is > we won't be able to answer the question experimentally. Those bothered > by possible loss of free will should recall that in a system complex > enough, there is room for the possibility of indeterminacy, be > it a biological system or whatnot. Not only is there room for the possibility of indeterminacy, some of us deliberately encorporate elements of randomness into our behaviors. (I for one don't want certain people to be able to predict every move I'm going to make.) As to Free Will, I define it as the capacity to make and enact choices consistent with my knowledge and values. > If humans are not a machine, what elements are added to the body > (which seems to be a physical machine as far as we can tell) > which make it otherwise? Are these material or immaterial? One of the more interesting elements that is added to the human body is the element of information. (There is at least one school of physics which proposes that the Universe is composed of matter, energy, and information.) The key information added to the human body is in the form of Knowledge and Values. In deference to our Eastern philosophical friends, we may think of such information as "Patterns of Organic Energy". (It is immaterial whether we think of such patterns as material or immaterial.) > Is there some aspect of human beings which does not obey the > laws of nature? Not to my knowledge. --Barry Kort
tmoody@sjuvax.UUCP (T. Moody) (12/08/88)
In article <1841@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >I will ask Serge the same questions I asked Gilbert: if humans are >not a machine, what elements are added to the body (which seems to >be a physical machine as far as we can tell) which make it otherwise? >Are these material or immaterial? Is there some aspect of human >beings which does not obey the laws of nature? The assumption here is that anything that "obeys the laws of nature" [as currently understood, or some future set?] is a machine. I have stayed out of the discussion so far, because this is a singularly uninteresting conception of "machine," in my view. If you don't understand "machine" in a way that lets you distinguish between, say, trees and clocks, then you are taking this word on a long holiday. -- Todd Moody * {allegra|astrovax|bpa|burdvax}!sjuvax!tmoody * SJU Phil. Dept. "The mind-forg'd manacles I hear." -- William Blake
throopw@xyzzy.UUCP (Wayne A. Throop) (12/09/88)
> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) > [...] most AIers believe the assertion that logic encapulates > the rules of thought, and that all sentences can be given a semantics > in formal logic (note how some famous mathematical logicians disagree > and stick to formal languages as being very different things). I dunno. I thought "most AIers" agreed with the formal logicians. It is true that "Knowledge Engineers" and "Expert System" designers are trying to model knowledge as imense sets of formal propositions. But it isn't clear to me that these constitute research so much as technologists spinning off results from former generations of AI research that didn't pan out *as* *AI* (as a whole) (but which might pan out as technology). > Anyone who talks of computers "understanding" does so: > a) to patronise users whom they don't know how to instruct properly; > b) because they are AI types. I dunno. The forms "the program knows <mumble>", or "the program will try to <mumble>" seem apt to me. No worse than "the moth knows its mate is upwind" or "the moth will try to reach its mate". I don't think these forms are necessarily silly anthropomorphisms. Hence, I don't think there is necessarily anything wrong with the AIer's use of "understanding", "know", "try" and so on as technical terms for what their programs are doing (or states that their programs are in). -- "Technical term" is a technical term for a common word or phrase whose meaning is, in some contexts, distorted beyond mortal comprehension. --- Hal Peterson cray!hrp -- Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw
peru@soleil.UUCP (Dave Peru) (12/09/88)
I was wrong. It's only a matter of time. Artificial Intelligence = Intelligence
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (12/09/88)
In article <1736@sjuvax.UUCP> tmoody@sjuvax.UUCP (T. Moody) writes: > >The assumption here is that anything that "obeys the laws of nature" [as >currently understood, or some future set?] is a machine. I have stayed >out of the discussion so far, because this is a singularly uninteresting >conception of "machine," in my view. If you don't understand "machine" >in a way that lets you distinguish between, say, trees and clocks, then >you are taking this word on a long holiday. Perhaps children would require such a restrictive concept of machine in order to differentiate trees and clocks, but I do not. I would be happy to hear of some other word, broad enough to include trees and clocks which we could use instead of machine. The concept as I am using it is that of a system which is potentially capable of being created, given sufficient natural (as opposed to supernatural) knowledge of its workings. The controversy is over whether humans (and I suppose plants and animals) are such systems. I hold that if humans are such "machines" then it is possible that someday we will be able to construct an artificial person.
cam@edai.ed.ac.uk (Chris Malcolm) (12/10/88)
In article <817@quintus.UUCP> ok@quintus.UUCP (Richard A. O'Keefe) writes: >In article <1841@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu >(Gordon E. Banks) writes: >> ..... >>The behaviorists have shown that behavior which subjectively seems >>to us to be caused by intention can be determined (even hypnotists >>can demonstrate this), > >Er, how do hypnotists demonstrate that? > .... >As for the first part of this, there is a philosophical tradition called >"compatibilism", which holds that "it was caused by intention" and >"it was determined" are not contradictory. I think what Gordon Banks is referring to is the rationalisation of post-hypnotic suggestion, where the victim is instructed under hypnosis to do something mildly bizarre at a certain time, and also to "forget" (have no conscious knowledge of) the instruction. At the appointed time the victim will perform the bizarre act, and on being asked why, will produce some spurious rationalisation, and insist under questioning that this rationalisation is the true, real, sincere motive of an act which was performed freely and with intention. Exactly the same phenomenon is exhibited by split-brain victims, when one half of the brain is asked to account for an action performed by the other half, where the action was taken on the basis of perceptual data only available to the performing half, and not to the explaining half. Once again, a spurious and often ingeniously contrived rationalisation is offered, with the speaker (speaking half) apparently quite sincere in believing it, and without any sensation of strain or puzzlement. A natural conclusion is that introspection is not a privileged window into the operations of the mind, but exactly the same kind of gifted hypothesizing we perform when "seeing" the motives of others as revealed by their behaviour in the context of our knowledge and suspicions, only of course with access to a larger fund of knowledge. In other words, the "subjective seeming" of one's introspection has exactly the same epistemic status as one's educated guess about the feelings, motivation, and mental processes of one's employer (for example). If this is true - and I think it is - then the commonsense folk psychology which is justified by appeals to our shared introspective experiences, although very useful for negotiating with one another, is dangerous mental luggage when doing AI research. Supposing free-will to be contradicted by determinism has always seemed to me to be due to confusing "determined" in the sense of "co-erced" (which is not free behaviour) with "determined" in the sense of "in principle predictable" which is quite another thing. Anyone who supposes that exercise of their free-will must depend upon some essentially unpredictable (random) component is clearly suffering from a shortage of good reasons for doing things, no? Chris Malcolm
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (12/10/88)
In article <42571@linus.UUCP> bwk@mbunix (Kort) writes: >Not only is there room for the possibility of indeterminacy, some >of us deliberately encorporate elements of randomness into our >behaviors. Interesting. Do you use a true or a pseudo random number generator to introduce this randomness? >One of the more interesting elements that is added to the human >body is the element of information. The information necessary to reproduce the organism seems to be encoded in DNA. One would suppose that knowledge and values are also encoded in physical systems whether neural networks or something as yet undiscovered. > (There is at least one >school of physics which proposes that the Universe is composed >of matter, energy, and information.) What school is this? I hadn't heard of it. >deference to our Eastern philosophical friends, we may think of >such information as "Patterns of Organic Energy". How does this Organic Energy differ from other forms of energy? Does it have separate conservation laws? Is this a new form of vitalism?
bwk@mitre-bedford.ARPA (Barry W. Kort) (12/10/88)
Richard A. O'Keefe says: > "An awareness of self" might be important to an intelligent organism, > but it might be a *consequence* of intelligence rather than a > *precondition* for it. The ability to repose a self model is a consequence of the ability to repose models. Model-based reasoning is one facet of intelligence, and a useful one for a sentient being who wishes to survive in a dangerous world. One of the interesting issues in defining a self model is the location of the boundary between the self and non-self portions of the model. There is some evidence that humans don't agree on the location of such boundaries. --Barry Kort
bwk@mitre-bedford.ARPA (Barry W. Kort) (12/11/88)
In article <1859@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) inquires about my earlier remarks: >In article <42571@linus.UUCP> bwk@mbunix (Kort) writes: >>Not only is there room for the possibility of indeterminacy, some >>of us deliberately encorporate elements of randomness into our >>behaviors. > >Interesting. Do you use a true or a pseudo random number generator >to introduce this randomness? Both, Gordon. For very long sequences, it becomes important if there is an adversary who is attempting to model my pattern of moves. >> (There is at least one >>school of physics which proposes that the Universe is composed >>of matter, energy, and information.) > >What school is this? I hadn't heard of it. I ran accros this item in the Science and the Citizen column of Scientific American some months ago. Perhaps another netter can provide a more detailed account. >>Indeference to our Eastern philosophical friends, we may think of >>such information as "Patterns of Organic Energy". > >How does this Organic Energy differ from other forms of energy? Does >it have separate conservation laws? Is this a new form of vitalism? It differs only in its structural relationship, the way pieces of a jigsaw puzzle relate to each other when they are assembled to form a big picture. Other than that, same pieces as the ones in the box. I am not familiar with the theory of vitalism. --Barry Kort
ok@quintus.uucp (Richard A. O'Keefe) (12/12/88)
In article <215@edai.ed.ac.uk> cam@edai (Chris Malcolm) writes: >In article <817@quintus.UUCP> ok@quintus.UUCP (Richard A. O'Keefe) writes: >>Er, how do hypnotists demonstrate that? >I think what Gordon Banks is referring to is the rationalisation of >post-hypnotic suggestion, where the victim is instructed under hypnosis >to do something mildly bizarre at a certain time, and also to "forget" >(have no conscious knowledge of) the instruction. At the appointed time >the victim will perform the bizarre act, and on being asked why, will >produce some spurious rationalisation, and insist under questioning >that this rationalisation is the true, real, sincere motive of an act >which was performed freely and with intention. There is a fairly major sort of non-sequitur here, plus a misapprehension of hypnotism. The non-sequitur is this: from: the act was suggested by the hypnotist the act was performed the actor produces a "spurious" rationalisation it does **NOT** follow that the act was not done freely. The misapprehension is the hypnotism is not a process whereby the hypnotist controls the will of the subject, but a voluntary fantasy which is particularly good at implanting false memories. For example, it is not the case that all people can be hypnotised, whereas if a subject can be hypnotised by one mesmerist he or she can usually be hypnotised by another. This ought to suggest to us that just maybe hypnosis might be something that subjects do, rather than something that hypnotists do. Let me propose another account of what might be going on. Subject agrees to play in hypnotic drama. Hypnotist makes suggestion. Subject voluntarily agrees to do so, rather than spoil the game. However, this doesn't seem like a good enough motive for the act, so subject confabulates another reason. (This is just cognitive dissonance at work.) Subject comes out of trance believing confabulation. Subject performs act. A key point here is cognitive dissonance. (Look it up in any good Psychology library.) People make up stories to account for their actions all the time, and believe them too. But it doesn't follow from that that their actions are not free.
fransvo@htsa (Frans van Otten) (12/12/88)
In article <1736@sjuvax.UUCP> tmoody@sjuvax.UUCP (T. Moody) writes: > If you don't understand "machine" >in a way that lets you distinguish between, say, trees and clocks, then >you are taking this word on a long holiday. When you are 'normally' talking about machines, you are right. But in this discussion, you are wrong. A 'biological system' (a human being, an animal, plants, trees, etc.) can be considered as black boxes. Put something into it (food), what happens ? (it grows). Now give information instead of food, what happens ? That's the way we are discussing human beings as machines. -- Frans van Otten Algemene Hogeschool Amsterdam Technische en Maritieme Faculteit fransvo@htsa.uucp
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (12/13/88)
From article <840@quintus.UUCP>, by ok@quintus.uucp (Richard A. O'Keefe): " ... For example, it is not the case that all people can be " hypnotised, whereas if a subject can be hypnotised by one mesmerist he or " she can usually be hypnotised by another. This ought to suggest to us that " just maybe hypnosis might be something that subjects do, rather than something " that hypnotists do. Some bottles are harder to open than others, therefore bottles open themselves. Some programs are harder to write than others, so computers have free will. (This method of argument has great possibilities.) " ... " A key point here is cognitive dissonance. (Look it up in any good " Psychology library.) People make up stories to account for their actions " all the time, and believe them too. ... All the time. Precisely. And one of the most popular is the story about free will. Greg, lee@uhccux.uhcc.hawaii.edu
ok@quintus.uucp (Richard A. O'Keefe) (12/13/88)
In article <2804@uhccux.uhcc.hawaii.edu> lee@uhccux.uhcc.hawaii.edu (Greg Lee) writes: >From article <840@quintus.UUCP>, by ok@quintus.uucp (Richard A. O'Keefe): >" ... For example, it is not the case that all people can be >" hypnotised, whereas if a subject can be hypnotised by one mesmerist he or >" she can usually be hypnotised by another. This ought to suggest to us that ^^^^^^^ >" just maybe hypnosis might be something that subjects do, rather than something ^^^^^ ^^^^^ >" that hypnotists do. >Some bottles are harder to open than others, therefore bottles open themselves. >Some programs are harder to write than others, so computers have free >will. (This method of argument has great possibilities.) I think you had better find some better analogies. At *NO* point in my message did I say ">therefore< people have free will". All I suggested in the passage quoted was that the fact that some people can be hypnotised and others can't **suggests** that hypnosis is something that the subjects do. THIS IS A TESTABLE HYPOTHESIS! (And in fact it is the current theory.) It does not follow from that that people have free will, nor that they don't. In this series of messages on this topic I have been careful to avoid stating my opinion about free will at all. I only criticsed a naive and out-of-date view of hypnosis. Some bottles are harder to open than others, this ought to suggest to us that a property of the bottle rather than the opener might be involved. Now _that_ would have been a fair analogy.
bwk@mitre-bedford.ARPA (Barry W. Kort) (12/13/88)
In article <215@edai.ed.ac.uk> cam@edai (Chris Malcolm) writes: > Anyone who supposes that exercise of their free-will must depend > upon some essentially unpredictable (random) component is clearly > suffering from a shortage of good reasons for doing things, no? I define Free Will as the capacity to make and enact choices consistent with my knowledge and values. It is only when my value system is teetering on the razor's edge between two choices that I turn to my random number generator to resolve the choice and get on with my life. --Barry Kort
smoliar@vaxa.isi.edu (Stephen Smoliar) (12/13/88)
In article <1859@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: > >> (There is at least one >>school of physics which proposes that the Universe is composed >>of matter, energy, and information.) > >What school is this? I hadn't heard of it. > The only proponent of this school with whom I am familiar is David Bohm, who gave two lectures on the subject at UCLA last year. The best place to find a summary of Bohm's ideas is in a book entitled UNFOLDING MEANING: A WEEKEND OF DIALOGUE WITH DAVID BOHM, edited by Donald Factor and published by Foundation House Publications in Gloucestershire.
pepke@loligo.fsu.edu (Eric Pepke) (12/13/88)
In article <42569@linus.UUCP> bwk@mbunix (Barry Kort) writes: >In article <180@loligo.fsu.edu> pepke@loligo.UUCP (Eric Pepke) writes: > > A random number generator can only be consulted a finite number of > > times in a lifetime. For every finite sequence of such random numbers, > > you can produce a partial Turing machine specification which produces > > that sequence. So, there's no problem. > >Just one problem, Eric. You have to build your Turing Machine >emulator before I have finished living my life. The information >you need to construct it is not available just yet. No, I just need to have it complete by the time you are ready to compare your life against it. Betcha that given enough resources I can make one between the time you die and the time you get used to flapping around your ectoplasm. >--Barry Kort -EMP "I don't know what this ectoplasm is that Arthur Conan Doyle keeps talking about, but it sounds like it would be great stuff to mend furniture with." -Archy
mintz@io.UUCP (Richard Mintz) (12/14/88)
In article <2221@xyzzy.UUCP> throopw@xyzzy.UUCP (Wayne A. Throop) writes: >> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) >> Anyone who talks of computers "understanding" does so: >> a) to patronise users whom they don't know how to instruct properly; >> b) because they are AI types. >I dunno. The forms "the program knows <mumble>", or "the program will >try to <mumble>" seem apt to me. No worse than "the moth knows its mate >is upwind" or "the moth will try to reach its mate". I don't think >these forms are necessarily silly anthropomorphisms. >Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw As a professional writer of user documentation, I heartily agree. Often the phrasing "if you do X, the software will do Y" is far preferable to the alternatives (usually "if you do X, you will see Y"). The former offers a more precise level of detail in an approximately equal number of words, with the added advantage of more active language. How (for example) can you concisely explain the steps in a complex software algorithm ("How a Document is Paginated") without resorting to constructions such as "First the software checks X, then it adjusts Y, then...."? The alternative ("The pagination algorithm consists of the following steps: [a] Verification of value X. [b] Adjustment of Y accordingly. [c]...") is likely to cause migraine headache in everyone except those who already understand, thus defeating the purpose of technical documentation. Rich Mintz eddie.mit.edu!ileaf!mintz The foregoing does not represent the opinion of Interleaf, Inc.
harper@hi-csc.UUCP (Paul L. Harper) (12/14/88)
I am continually amazed at the faith of AI "researchers" (programmers?). I have seen nothing whatsoever from the AI community that indicates there is any hope of producing intelligence by running instructions on computers. It is an incredible leap of faith, completely unfounded by science, to assume that computers can obtain the human quality we call intelligence. Where is the scientific justification for the assumption?
peru@soleil.UUCP (Dave Peru) (12/15/88)
>>In article <215@edai.ed.ac.uk> cam@edai (Chris Malcolm) writes: >> >> Anyone who supposes that exercise of their free-will must depend >> upon some essentially unpredictable (random) component is clearly >> suffering from a shortage of good reasons for doing things, no? In article <42939@linus.UUCP> (Barry W. Kort) writes: >I define Free Will as the capacity to make and enact choices consistent >with my knowledge and values. It is only when my value system is >teetering on the razor's edge between two choices that I turn to my >random number generator to resolve the choice and get on with my life. Sometimes I think my brain is a chunk of clay. Incoming reality molds and defines my knowledge and values. Sometimes I wonder whether I or anyone else has Free Will at all. Just think, if the universe is deterministic, at the time of the Big-Bang, you could have predicted this article and the exact words used. Remember when a person could get a 30 year fixed rate mortgage at 7%. Cigarette smoke is toxic waste, do you have a choice?
andy@cs.columbia.edu (Andy Lowry) (12/15/88)
In article <4040a289.9d8d@hi-csc.UUCP> harper@hi-csc.UUCP (Paul L. Harper) writes: >I am continually amazed at the faith of AI "researchers" >(programmers?). I have seen nothing whatsoever from the AI >community that indicates there is any hope of producing >intelligence by running instructions on computers. > >It is an incredible leap of faith, completely unfounded >by science, to assume that computers can obtain the human >quality we call intelligence. Where is the scientific justification >for the assumption? I am continually amazed at the closed-mindedness of certain individuals. On the contrary... it is an incredible leap of faith (in my book) to assume that this goal is unattainable. That is, I cannot conceive of ANY argument that intelligence cannot be fabricated other than one based on a belief in God. And that is a belief that I do not hold any part of, and that I consider an "incredible leap of faith." That I believe "true" artificial intelligence to be attainable does not mean that I necessarily believe it will be attained. That depends on a fair amount of luck, among other things. It does mean that I consider it a worthy goal for research effort. In fact, even if I were not so convinced that the goal can, theoretically, be achieved, I would still consider it a worthy pursuit. How many programs of research are undertaken with full confidence in their eventual success? Attempting to obtain a goal is certainly one valid way to go about seeing how attainable it is. -Andy
josh@klaatu.rutgers.edu (J Storrs Hall) (12/16/88)
Paul Harper writes:
I am continually amazed at the faith of AI "researchers"
(programmers?). I have seen nothing whatsoever from the AI
community that indicates there is any hope of producing
intelligence by running instructions on computers.
Why "continually"? If this is so off-the-wall a pursuit, why does Mr.
Harper bother "continually" reading this newsgroup, and worrying about
it? Why doesn't he read talk.politics instead, and be continually
amazed that there are so many people that believe any of the creeds
expounded there?
It is an incredible leap of faith, completely unfounded
by science, to assume that computers can obtain the human
quality we call intelligence. Where is the scientific justification
for the assumption?
The scientific method has little to say about justification for
assumptions. AI is a scientific hypothesis, a theory. AI
practitioners are doing experiments, or if you wish, contributing
to the big overall experiment, to test the theory.
There is an attitude, all too common among business DP types,
that "If I can't program it in COBOL, it's worthless." (For
science/engineering types, /COBOL/FORTRAN/.) It turns out that
there are plenty are reasons to believe that (a) intelligence
is possible in computers of sufficient memory and processing
power; (b) it will be possible for us to create such intelligent
programs after a sufficient investment in software capital; and
(c) the knowledge gained in attempting this will be interesting
and useful.
However, it would be boring and worthless to explain this to
Mr. Harper, so I shall not try. Instead I will merely offer him
the advice that if AI baffles him, ignore it, and we'll all be
better off.
--JoSHbph@buengc.BU.EDU (Blair P. Houghton) (12/16/88)
In article <4040a289.9d8d@hi-csc.UUCP> harper@hi-csc.UUCP (Paul L. Harper) writes: >I am continually amazed at the faith of AI "researchers" >(programmers?). I have seen nothing whatsoever from the AI >community that indicates there is any hope of producing >intelligence by running instructions on computers. > >It is an incredible leap of faith, completely unfounded >by science, to assume that computers can obtain the human >quality we call intelligence. Where is the scientific justification >for the assumption? In a hundred and fifty years of neural science that has determined the primary functions of the elements involved in thought and has but to determine the architecture before it understands the function of the whole machine. This member of the AI community says: "give me a map of the brain and I'll make it compose piano sonatas while solving the middle-east peace problem, and I'll do it all on my little Connection Machine. Just don't expect it to do it quickly, and don't expect it this century. A trillion neurons is a lot of code." --Blair "Expert System is an oxymoron."
sworking@teknowledge-vaxc.ARPA (Scott Workinger) (12/16/88)
In article <42571@linus.UUCP> bwk@mbunix (Kort) writes: >(There is at least one >school of physics which proposes that the Universe is composed >of matter, energy, and information.) Having come to the same conclusion, independently, I would be very interested in seeing a reference on this. (Another way of looking at it is that everything that exists is categorizable as matter, energy, and structure, where structure is equivalent to information.) Some people of a more philosophical bent would say that it is the hole in the center of the wheel that makes it useful. Scott
throopw@xyzzy.UUCP (Wayne A. Throop) (12/17/88)
> tmoody@sjuvax.UUCP (T. Moody) > If you don't understand "machine" > in a way that lets you distinguish between, say, trees and clocks, then > you are taking this word on a long holiday. This is a little unclear to me. Do you mean that not to place a boundary on the definition of "machine" between trees and clocks is wrong-headed, or do you mean that the definition of machine must be segmented in such a way that a boundary (of what fuzziness?) must exist between trees and clocks? But question: *Is* there any difference in the way trees and clocks operate (except for the obvious difference in complexity)? And further: what about between trees and humans? -- The real problem is not whether machines think, but whether men do. --- B.F. Skinner -- Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (12/17/88)
From article <26161@teknowledge-vaxc.ARPA>, by sworking@teknowledge-vaxc.ARPA (Scott Workinger):
" In article <42571@linus.UUCP> bwk@mbunix (Kort) writes:
"
" >(There is at least one
" >school of physics which proposes that the Universe is composed
" >of matter, energy, and information.)
"
" Having come to the same conclusion, independently, I would be
" very interested in seeing a reference on this. (Another
I have seen something by Erwin Schro"dinger sort of in this
direction. Sorry I can't give a specific reference.
Greg, lee@uhccux.uhcc.hawaii.edusbigham@dukeac.UUCP (Scott Bigham) (12/17/88)
In article <Dec.15.11.41.17.1988.12131@klaatu.rutgers.edu> josh@klaatu.rutgers.edu (J Storrs Hall) writes: > It turns out that >there are plenty are reasons to believe that (a) intelligence >is possible in computers of sufficient memory and processing >power; (b) it will be possible for us to create such intelligent >programs after a sufficient investment in software capital; and >(c) the knowledge gained in attempting this will be interesting >and useful. What are they? I'd love to know, because I'm inclined not to believe any of the above (except for (c), which I firmly support). Inquiring minds wanna know (computers rarely if ever inquire...) sbigham -- Scott Bigham "The opinions expressed above are Internet sbigham@dukeac.ac.duke.edu (c) 1988 Hacker Ltd. and cannot be USENET sbigham@dukeac.UUCP copied or distributed without a ...!mcnc!ecsgate!dukeac!sbigham Darn Good Reason."
marty@homxc.UUCP (M.B.BRILLIANT) (12/17/88)
In article <4040a289.9d8d@hi-csc.UUCP>, harper@hi-csc.UUCP (Paul L. Harper) drew a lot of FLAK for writing: > I am continually amazed at the faith of AI "researchers" > (programmers?). I have seen nothing whatsoever from the AI > community that indicates there is any hope of producing > intelligence by running instructions on computers. > > It is an incredible leap of faith, completely unfounded > by science, to assume that computers can obtain the human > quality we call intelligence. Where is the scientific justification > for the assumption? The probable reason why he drew so much FLAK is that there is some truth in what he said, some elements that are unfair, and much that is unwelcome and confusing to some people. It is a matter of faith to some that the activities of living beings are produced by the interaction of objectively detectable matter and energy, and therefore technology can come progressively closer to understanding and reproducing them. It is a matter of faith to others that the same activities of living beings depend on something that is not explainable by the interaction of objectively detectable matter and energy, and therefore technology will always fall short of understanding and reproducing them. Two points in the preceding statements are worth noting: First, the philosophical bases are diametrically opposite. Second, the conclusions are totally consistent and cannot be distinguished by observation. To a perfect scientist or engineer, it does not matter whether you say that we will come ever closer to simulating life, or that we will never perfectly simulate life. Both are good working hypotheses. Once we realize that the argument is essentially religious, we will stop arguing about it. M. B. Brilliant Marty AT&T-BL HO 3D-520 (201) 949-1858 Home (201) 946-8147 Holmdel, NJ 07733 att!houdi!marty1 Disclaimer: Opinions stated herein are mine unless and until my employer explicitly claims them; then I lose all rights to them.
josh@klaatu.rutgers.edu (J Storrs Hall) (12/17/88)
Marty Brilliant writes: In article <4040a289.9d8d@hi-csc.UUCP>, harper@hi-csc.UUCP (Paul L. Harper) drew a lot of FLAK for writing: > I am continually amazed at the faith of AI "researchers" ... The probable reason why he drew so much FLAK is that there is some truth in what he said, some elements that are unfair, and much that is unwelcome and confusing to some people. The real reason he drew so much "FLAK" is that he wanted to, and said exactly what would press the most buttons on an AI devotee. Even if every word he said were true, and AI really a "leap of faith" religious creed, his action in posting would be equivalent to running into a church service and shouting denunciations of the worshippers. If AI is a religion, why not let us practice it in peace? --and of course if it is not, the denunciations are stupid as well as malicious. --JoSH
jeff@aipna.ed.ac.uk (Jeff Dalton) (12/18/88)
In article <4040a289.9d8d@hi-csc.UUCP> harper@hi-csc.UUCP (Paul L. Harper) writes: >It is an incredible leap of faith, completely unfounded >by science, to assume that computers can obtain the human >quality we call intelligence. Where is the scientific justification >for the assumption? Where is anyone assuming this? People are (1) trying to see what machines can do, (2) wondering whether they might be able to do sufficiently complex, etc. things so as to be considered intelligent, (3) noticing that computers are pretty general beasts and that so far we haven't found any hard limits that say they must fall short of intelligence, (4) pointing out what they think are mistakes in arguments that AI won't work. Why should no one ever try to make machines that can identify objects in images or perform other new and interesting tasks until they are sure it could be done? Very little AI work is directed at "real intelligence". It's clear that we're very far from that and not clear that it's even possible. But artificial insects, for example, might be another matter.
rjc@aipna.ed.ac.uk (Richard Caley) (12/18/88)
In article <4040a289.9d8d@hi-csc.UUCP> harper@hi-csc.UUCP (Paul L. Harper) writes: >It is an incredible leap of faith, completely unfounded >by science, to assume that computers can obtain the human >quality we call intelligence. Where is the scientific justification >for the assumption? Where is the scientific justification for the oposite assumption? It is an open question, and one certainly gets no answers by assuming it is impossible, so the best strategy seems to be to assume it is possible and try to disprove this, since this could be done by one facet of human "inteligence" which is proved to be incomputable. There is no "scientific" justification for the assumption that the entire universe obeys those laws we have observed locally on the earth, it is an assumption. One must assume something to get anywhere. -- rjc@uk.ac.ed.aipna AKA rjc%uk.ac.ed.aipna@nss.cs.ucl.ac.uk "We must retain the ability to strike deep into the heart of Edinburgh" - MoD
dc@gcm (Dave Caswell) (12/19/88)
In article <4639@homxc.UUCP> marty@homxc.UUCP (M.B.BRILLIANT) writes:
(ONE)
.It is a matter of faith to some that the activities of living beings
.are produced by the interaction of objectively detectable matter and
.energy, and therefore technology can come progressively closer to
.understanding and reproducing them.
(TWO)
.It is a matter of faith to others that the same activities of living
.beings depend on something that is not explainable by the interaction
.of objectively detectable matter and energy, and therefore technology
.will always fall short of understanding and reproducing them.
Is it safe to say that the people who don't think there is any faith
involved in believing (ONE) think that it is at least possible we
shall someday attain machine intelligence? Is there any reason to
think that believing in (ONE) is an "act of faith"? Is there any
reason to believe in (TWO) except faith? My mind is made up; I
wouldn't have used the words "a matter of faith" in (ONE).
--
Dave Caswell
Greenwich Capital Markets uunet!philabs!gcm!dcfransvo@htsa.uucp (Frans van Otten) (12/19/88)
In article <2804@uhccux.uhcc.hawaii.edu> lee@uhccux.uhcc.hawaii.edu (Greg Lee) writes: >From article <840@quintus.UUCP>, by ok@quintus.uucp (Richard A. O'Keefe): >" ... For example, it is not the case that all people can be >" hypnotised, whereas if a subject can be hypnotised by one mesmerist he or >" she can usually be hypnotised by another. This ought to suggest to us that >" just maybe hypnosis might be something that subjects do, rather than something >" that hypnotists do. > >Some bottles are harder to open than others, therefore bottles open themselves. >Some programs are harder to write than others, so computers have free >will. (This method of argument has great possibilities.) > > Greg, lee@uhccux.uhcc.hawaii.edu Let's define hypnosis as 'being in an altered state of consciousness'. Then the question becomes: how does someone get into this state ? There are some possibilities: 1) you do it yourself, 2) you agree with someone that (s)he helps you, 3) someone having 'hypnotizing power' 'does it to you' before you are aware it's happened. Note on 3): In my opinion, you must have some vulnerable spot in your per- sonality which the other person (knowing it or not) uses. I have never met someone without such a spot. By the way, the hypnotizing person doesn't need to be aware of what (s)he is doing; (s)he may very well be doing this unconsciously (because (s)he has some psycholocal trauma/.../...) Also, from personal experience I believe that everyone can be hypnotized. It is true that this is very hard sometimes; even someone who wants to get hypnotized may have blocks on other (unconscious/subconscious) levels. -- Frans van Otten Algemene Hogeschool Amsterdam Technische en Maritieme Faculteit fransvo@htsa.uucp
fransvo@htsa.uucp (Frans van Otten) (12/19/88)
In article <4040a289.9d8d@hi-csc.UUCP> harper@hi-csc.UUCP (Paul L. Harper) writes: >I am continually amazed at the faith of AI "researchers" >(programmers?). I have seen nothing whatsoever from the AI >community that indicates there is any hope of producing >intelligence by running instructions on computers. > >It is an incredible leap of faith, completely unfounded >by science, to assume that computers can obtain the human >quality we call intelligence. Where is the scientific justification >for the assumption? Where is the (scientific ?) justification for the above assumption ? -- Frans van Otten Algemene Hogeschool Amsterdam Technische en Maritieme Faculteit fransvo@htsa.uucp
markh@csd4.milw.wisc.edu (Mark William Hopkins) (12/21/88)
In article <17@ubc-cs.UUCP> morrison@grads.cs.ubc.ca (Rick Morrison) writes: >In article <4264@homxc.UUCP> marty@homxc.UUCP (M.B.BRILLIANT) writes: >> ... Do we know what ``artificial intelligence'' is? Or are we just talking >> about something we don't know anything about? ... Just somebody >>tell us what the answer is. > >The answer is "who cares?" Does anyone in this group actually _do_ AI? Well, if Natural Language processing counts as AI, the answer is yes. >I'm beginning to think that the most appropriate definition of AI is >"a discipline concerned with the uninformed examination of unresolvable >philosophical and psychological issues." I do believe that human intelligence is beyond human comprehension. The reason is that the day we learn about our intelligence as it currently exists, we'll experience a quantum leap in our own intelligence AS A RESULT. So we'll always be one step behind ourselves. None of this rules out AI, though ... especially if AI were to be automated ;-). Another, related, posting described the impossibility of AI by using the classic Sentimentalist argument: machines won't be able to recognize a Duck when it sees one so it could "throw" it a piece of bread out of humanitarian concern. The machine's response: "I see no cause for that kind of insult! Really! Comparing *ME* to a human!" Signals in our nervous system travel at about 700 MPH (if my memory is correct). Signals in Silicon travel about 1 *MILLION* times faster. It's not whether AI is possible, no, the question is how long it will be before the machine's capacity exceeds our own, as it will. Being scared of the possibility is no excuse for denying its inevitability, because the machine is our own creation and our own tool, albeit an intelligent one. But we don't deify cars because they travel faster than us (well, normal people don't, at least), so nobody would ever deify a human artifact that happened to think faster and better than us. And as for unresolvable philosophical problems such as the one behind Goedel's Theorem: you may not realise it, but even there there is a tiny crack that may allow for a resolution. You see, nobody ever showed that the negation of the key statement which Goedel used in his theorem is not proveable. In fact, the Number Theory that Goedel looked at may actually be both complete and consistent if that negation IS proveable within the original theory itself. What Number Theory would be then is "omega-consistent" ... in which case you would have infinite numbers, infinitesimals and the like rammed down your throat ... a possibility I savor, and a perfect resolution to the philosophical dilemma.
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (12/22/88)
In article <4040a289.9d8d@hi-csc.UUCP> harper@hi-csc.UUCP (Paul L. Harper) writes: >I am continually amazed at the faith of AI "researchers" >(programmers?). I have seen nothing whatsoever from the AI >community that indicates there is any hope of producing >intelligence by running instructions on computers. > >It is an incredible leap of faith, completely unfounded >by science, to assume that computers can obtain the human >quality we call intelligence. Where is the scientific justification >for the assumption? Let me try to briefly answer your question. The assumption derives from the following assumptions: 1. Intelligence is a function of the human brain. 2. The brain is a physical object and its functioning is explainable in terms of its organization and the laws of physics. 3. Given sufficient understanding of the composition and organization of the brain, and sufficient progress in technology, it should be possible to artificially create intelligence (or for Gilbert Cockton, a system capable of acquiring intelligence through appropriate socialization). As far as creating a complete intelligence by "executing instructions", I would have to know what you mean by that term. If you are talking about a Turing machine, then I would say that if you had one fast enough, you could probably simulate all of the functions of the massively parallel brain serially and create such an intelligence, but I seriously doubt if that is the way it will be done (at least at first), since that would require more knowledge than other approaches. Of course if you do not accept the 3 assumptions (for example, if you believe that intelligence is a function of man's spirit rather than the brain) then you have a logical reason for rejecting the idea of an artificial intelligence. (Even that objection might be met with the notion that given a sufficiently complex machine, a spirit might be found to inhabit it, as well, although that is truly a leap of faith!)
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (12/22/88)
In article <4639@homxc.UUCP> marty@homxc.UUCP (M.B.BRILLIANT) writes: > >It is a matter of faith to others that the same activities of living >beings depend on something that is not explainable by the interaction >of objectively detectable matter and energy, and therefore technology >will always fall short of understanding and reproducing them. > >Two points in the preceding statements are worth noting: > >First, the philosophical bases are diametrically opposite. > >Second, the conclusions are totally consistent and cannot be >distinguished by observation. > Perhaps I can agree with your first statement but not your second. If indeed an artificial intelligence is created, that action will negate the premise. Almost all unknown phenomena were initially given supernatural explanations. The history of science traces out a point by point collapse of such notions (although there are still stragglers who refuse to give up even the notion of the flat earth). I feel that it would be better if people would avoid staking their religious beliefs on such questions, but I am sure religion will survive the creation of artificial persons. Wouldn't it be interesting if these artificial persons were also religious? (Maybe I'll write a SF story abou this.)
bickel@nprdc.arpa (Steven Bickel) (12/23/88)
In article <1901@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: > {...} for example, if you >believe that intelligence is a function of man's spirit rather than >the brain) then you have a logical reason for rejecting the idea >of an artificial intelligence. (Even that objection might be met >with the notion that given a sufficiently complex machine, a spirit >might be found to inhabit it, as well, although that is truly a leap >of faith!) An interesting twist that I have often pondered is that intelligent life might require a machine (brain etc.) within an energy field (ie. electromagnetic or some derivative that we can or cannot currently measure). Just a metaphysical thought. :-) Steve Bickel
anderson@secd.cs.umd.edu (Gary Anderson) (12/23/88)
The successful application of artificial intelligence tools
and techniques will increase the degree to which humans depend
on machines. Clearly, human kind will be able to construct
smart, productive, efficient machines to carry out many important tasks.
These machines will become superior to humans
along certain dimensions (ie computational speed, physical strength, agility,
stamina) in specific applications. As the performance of these machines
improves, and their use becomes more pervasive, we will come to
rely on them more and more for our survival and comfort.
These ubiquitous machines will touch the lives of nearly everyone and
they will be a source of great anxiety for many people.
As these machines become more complex and more versatile,
their behavior in new and unanticipated situations will become more
difficult to predict. Even hindsight understanding of the behavior
of these machines in new situations may be very difficult to achieve.
I wonder if even the developers of these machines will be able to
distinguish free will from poorly understood programmed behavior.
If the machines become so smart and so complex that we cannot
easily predict their behavior in new situations, we
will have no recourse but to ask "them" what they would do.
It is my impression that a major reason for hoping to observe consciousness
and free will in our machines is that observing these anthropomorphic
characteristics would ease our understandable anxiety about depending on
the behavior of machines we don't fully understand.
Consciousness would provide another
channel for communicating with and "understanding"
( or at least *ir*rationalizing :-} )
the behavior of these complicated machines.
I wonder how effective this channel would be in reducing anxiety about
smart machines, and how useful it would be in predicting, monitoring
and controlling their behavior. The results are mixed for human-human
exploitation of this channel. Additionally, I think there would be some
difficult moral/ethical issues associated with manipulating the actions of
an agent with "free will".
--
Gary S. Anderson | Probity, sincerity, candor,
| conviction, the idea of duty,
+-+-+-+-+-+-+-+-+-+-+-+-+ | are things which, mistaken, may
email: anderson@secd.cs.umd.edu | become hideous, but which even
U.S. Snail: University of Maryland | though hideous, remain great;
Department of Economics | their majesty, peculiar to the
Room 3147c Tydings Hall | human conscience, continues in
College Park, MD 20742 | all their horror; they are
Voice: (301)-454-6356 | virtues with a single vice ---
| error.
---------------------------------------------- Victor Hugo, Les Miserablesned@h-three.UUCP (ned) (12/24/88)
In article <16@csd4.milw.wisc.edu>, markh@csd4.milw.wisc.edu (Mark William Hopkins) writes: > I do believe that human intelligence is beyond human comprehension. The > reason is that the day we learn about our intelligence as it currently > exists, we'll experience a quantum leap in our own intelligence AS A RESULT. > So we'll always be one step behind ourselves. I don't think that the *mechanics* of human intelligence are beyond human comprehension or will change due to any increase in knowledge. Of course, the brain will evolve, but I think we have time to figure it out before our information becomes obsolete. :-) > Signals in our nervous system travel at about 700 MPH (if my memory is > correct). Signals in Silicon travel about 1 *MILLION* times faster. > It's not whether AI is possible, no, the question is how long it will be > before the machine's capacity exceeds our own, as it will. What if it turns out that chemical processes are practically the only means of creating human-like intelligence? If so, our machines may not be any faster or more capable than the brain. -- Ned Robie uunet!h-three!ned
harper@hi-csc.UUCP (Paul L. Harper) (12/27/88)
(Andy Lowry @ Columbia University Department of Computer Science) writes: > In article <4040a289.9d8d@hi-csc.UUCP> harper@hi-csc.UUCP (Paul L. Harper) writes: > >I am continually amazed at the faith of AI "researchers" > >(programmers?). I have seen nothing whatsoever from the AI > >community that indicates there is any hope of producing > >intelligence by running instructions on computers. > > > >It is an incredible leap of faith, completely unfounded > >by science, to assume that computers can obtain the human > >quality we call intelligence. Where is the scientific justification > >for the assumption? > > I am continually amazed at the closed-mindedness of certain > individuals. On the contrary... it is an incredible leap of faith (in > my book) to assume that this goal is unattainable. That is, I cannot > conceive of ANY argument that intelligence cannot be fabricated other > than one based on a belief in God. And that is a belief that I do not > hold any part of, and that I consider an "incredible leap of faith." > > That I believe "true" artificial intelligence to be attainable does > not mean that I necessarily believe it will be attained. That depends > on a fair amount of luck, among other things. It does mean that I > consider it a worthy goal for research effort. > > In fact, even if I were not so convinced that the goal can, > theoretically, be achieved, I would still consider it a worthy > pursuit. How many programs of research are undertaken with full > confidence in their eventual success? Attempting to obtain a goal is > certainly one valid way to go about seeing how attainable it is. It is interesting that the response to my posting makes no attempt at answering the major query, i.e. what about scientific justification? The justification is exactly that which I have complained about, being "I have faith". I will grant that the feeling that "I'm on the right track" or something similar is viable in scientific pursuits, especially for us humans. But after so many years of AI promises, little of consequence seems to have been produced. A couple of questions: What is ' "true" artificial intelligence ' ? Will "true" artificial intelligence have consciousness? Paul
harper@hi-csc.UUCP (Paul L. Harper) (12/28/88)
J Storrs Hall @ Rutgers Univ., New Brunswick, N.J. writes (in response to my posting): > It turns out that > there are plenty are reasons to believe that (a) intelligence > is possible in computers of sufficient memory and processing > power; (b) it will be possible for us to create such intelligent > programs after a sufficient investment in software capital; and > (c) the knowledge gained in attempting this will be interesting > and useful. > > However, it would be boring and worthless to explain this to > Mr. Harper, so I shall not try. Instead I will merely offer him > the advice that if AI baffles him, ignore it, and we'll all be > better off. Whew! Rather than respond to the tone of the above, I'd like to ask what the reasons are for believing (a) and (b) above. What are the foundations for believing in the above? Is this based on a belief in functional AI (if the functions that make up intelligence can be identified, in terms of input and output relationships, then any implementation is adequate), or does the requirement of memory and processing power stem from the belief that simulating the brain (to presumably a rather fine level of detail) would be a means to attain AI? I am not *attacking* AI; I'm just looking for scientific justification for the many claims, etc. Paul
bwk@mbunix.mitre.org (Barry W. Kort) (12/28/88)
In article <1904@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) injects in interesting thought: > ... I am sure religion will survive the creation > of artificial persons. Wouldn't it be interesting if these artificial > persons were also religious? (Maybe I'll write a SF story about this.) If artificial persons exhibit the property of seeking new knowledge, then they will profess a belief system which motivates their behavior. They will profess the belief that new knowledge awaits their discovery, and that puzzles and unsolved problems will succumb to thoughtful lines of reasoning and meticulous methods of investigation. Perhaps we should coin a name for this system of religious belief. I propose we call it Science. --Barry Kort
bwk@mbunix.mitre.org (Barry W. Kort) (12/28/88)
In article <15152@mimsy.UUCP> anderson@secd.cs.umd.edu (Gary Anderson speculates about future generations of thinking machines: > As these machines become more complex and more versatile, > their behavior in new and unanticipated situations will become more > difficult to predict. Even hindsight understanding of the behavior > of these machines in new situations may be very difficult to achieve. > I wonder if even the developers of these machines will be able to > distinguish free will from poorly understood programmed behavior. > If the machines become so smart and so complex that we cannot > easily predict their behavior in new situations, we > will have no recourse but to ask "them" what they would do. I suspect the distinction between programmed behavior and free will will become a fuzzy boundary. A chess playing computer, when confronted with a novel situation may choose at random from a small set of alternatives. The outcome (win or lose) may then become compiled knowledge about the wisdom of the chosen line of play. The next time around, the chess machine won't be so naive, and may choose it's course of action with more conviction. The bemused observer would be hard pressed to distinguish free will from such random decision. So, asking them what they would do in a hypothetical situation might generate the honest answer, "I don't know. It depends on whether I learn something useful by the time I have to make that decision." --Barry Kort
bwk@mbunix.mitre.org (Barry W. Kort) (12/31/88)
In article <4081e1ba.75f0@hi-csc.UUCP> harper@hi-csc.UUCP (Paul L. Harper) asks: > What is ' "true" artificial intelligence ' ? > Will "true" artificial intelligence have consciousness? I define true artificial intelligence as intelligence residing on a substrate other than a biologically grown carbon-based neural network. (Possibly a silicon substrate fabricated with currently available technology.) I believe that an intelligent system which pursues the goal of knowledge acquisition over time will exhibit behavior indistinguishable from that of a conscious sentient being with transient emotional states. --Barry Kort
peru@soleil.UUCP (Dave Peru) (01/04/89)
"But essential abilities for intelligence are certainly:
to respond to situations very flexibly;
to take advantage of fortuitous circumstances;
to make sense out of ambiguous or contradictory messages;
to recognize the relative importance of different elements of a situation;
to find similarities between situations despite differences which may
separate them;
to draw distinctions between situations despite similarities which may
link them;
to synthesize new concepts by taking old concepts and putting them
together in new ways;
to come up with ideas with are novel.
Here one runs up against a seeming paradox. Computers by their very nature
are the most inflexible, desireless, rule-following of beasts. Fast though
they may be, they are nonetheless the epitome of unconsciousness. How, then,
can intelligent behavior be programmed? Isn't this the most blatent of
contradictions in terms? One of the major theses os this book is that it is
not a contradiction at all. One of the major purposes of this book is to urge
each reader to confront the apparent contradiction head on, to savor it, to
turn it over, to take it apart, to wallow in it, so that in the end the reader
might emerge with new insights into the seemingly unbreachable gulf between
the formal and the informal, the animate and the inanimate, the flexible and
the inflexible.
This is what Artificial Intelligence (AI) research is all about. And the
strange flavor of AI work is that people try to put together long sets of
rules in strict formalisms which tell inflexible machines how to be flexible."
Taken from p.26 of "Goedel, Escher, Bach..." by D.R. Hofstadter.
Does anyone disagree with this?
Does anyone strongly disagree if I include in the definition of "intelligence"
the ability to recognize a paradox?fransvo@htsa.uucp (Frans van Otten) (01/05/89)
In article <552@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: >"But essential abilities for intelligence are certainly: > > to respond to situations very flexibly; > to take advantage of fortuitous circumstances; > to make sense out of ambiguous or contradictory messages; [ etc. ] >Does anyone disagree with this? I disagree. This is not a definition of intelligence, nor are the listed abilities essential for intelligence. These are mere examples. I really stick to my definition of intelligence: *** Intelligence: The ability to draw a conclusion. *** Needed: A database and an algorithm to reach a conclusion based on the data. *** Improvements: The ability to change the database. The conclusion-algorithm being part of the database, so that the system can add/change algorithms. I would like to know how other people think about my definition. -- Frans van Otten Algemene Hogeschool Amsterdam Technische en Maritieme Faculteit fransvo@htsa.uucp
peru@soleil.UUCP (Dave Peru) (01/07/89)
Please consider the following thoughts of three people concerning the physics
of the mind. Notice the difference from the first person and the next two.
COMPUTER SCIENTIST:
In the book "The Society of Mind" Marvin Minsky writes (p.50):
"When people have no answers to important questions, they often give some
anyway.
What controls the brain? The Mind.
What controls the mind? The Self.
What controls the Self? Itself.
To help us think about how our minds are connected to the outer world, our
culture teaches schemes like this:
(diagram ...)
This diagram depicts our sensory machinery as sending information to the
brain, wherein it is projected on some inner mental movie screen. Then,
inside that ghostly theater, a lurking Self observes the scene and then
considers what to do. Finally, that Self may act--somehow reversing all
those steps--to influence the real world by sending various signals back
through yet another family of remote-control accessories.
This concept simply doesn't work. It cannot help for you to think that
inside yourself lies someone else who does your work. This notion of
"hommunculus"--a little person inside each self--leads only to a paradox
since, then, that inner Self requires yet another movie screen inside itself,
on which to project what *it* has seen! And then, to watch that
play-within-a-play, we'd need yet another Self-inside-a-Self--to do the
thinking for the last. And then this would all repeat again, as each new
Self requires yet another one to do its job!
The idea of a single, central Self doesn't explain anything. This
is because a thing with no parts provides nothing that we can use
as pieces of explanation!
Then why do we so often embrace the strange idea that what we do is done
by Someone Else--that is, our Self? Because so much of what our minds
do is hidden from the parts of us that are involved with verbal
consciousness."
MATHEMATICIAN/PHYSICIST/ASTRONOMY:
In the book "Bridges To Infinity" Michael Guillen (Ph.D in physics, mathema-
matics, and astronomy from Cornell University) writes (p.98):
"In his thirteen-page manuscript, "All Numbers, Great and Small," Conway
begins as Frege began, with a few primitive ideas, including the null set
and two rules. The first rule, Conway's logical definition of a number,
can be visualized in terms of encyclopedia volumes lined up in order in
a library shelf. According to the definition, a volume's place in the
lineup, its number, can be inferred from the set of volumes on its left
and the set of volumes on its right. We could determine where volume nine
belongs, for instance, simply by locating that place where volumes zero
through eight are on the left and volumes ten through infinity are on the
right. Therefore, every volume, every number, has its own niche, determined
uniquely by the left and right sets. That's the thrust of Conway's first
rule.
His second rule, again explained here in terms of a set of encyclopedias,
decrees that one number, such as 5, is smaller than (or equal to) another
number, such as 9, if two things are true simultaneously: (A) all the volumes
to the left of the first number (5) are less than the second number (9),
and (B) all the volumes to the right of the second number (9) are bigger
than the first number (5). This rule is necessary in order for Conway
to impose an order on the numbers he creates, beginning with zero: Zero
is less than 1, so it precedes 1; 1 is less than 2, so it precedes 2; and
so forth.
As he does not assume the existence of any numbers to begin with, Conway,
like Frege, has only the null set with which to start creating the sequence
of natural numbers. Consequently, Conway first contemplates the number
whos left and right sets are both null sets, written symbolically as {}:{},
He names this *zero*. That is, in Conway's theory, as in Frege's, nothingness
is the most primitive realization of nothing.
After creating the number zero, Conway has two sets with which to continue
creating numbers: the null set, {}, and the set containing zero, {0}.
Conway identifies the number 1 as the number whose left set contains zero
and whose right set is the null set. Thus, at this point in Conway's genesis,
the number 1 is flanked to the left by nothingness and to the right by
nothing. To the left is potential already realized (as zero), and to the
right is potential not yet realized.
At each point in his creation, Conway always selects the next number as
the number as the number whose left set contains all the previously created
numbers and whose right set is the null set. It's as though he were being
guided by an image of those encyclopedias. At each point, the newly created
volume is placed to the right of all those volumes already shelved and
to the left of empty space, which in this analogy has the aspect of the
physicist's vacuum in representing the potential of numbers not yet brought
into being. By proceeding in this fashion indefinitely, Conway creates
the entire sequence of natural numbers.
From there he goes on, however, to create an infinity of in-between numbers,
such as the number whose left set contains zero, {0}, and whose right set
contains one through infinity {1, 2, 3, ...}. This defines a number somewhere
between zero and one. Thus the standard set of encyclopedias, the natural
numbers, is embellished by an interminable number of in-between volumes.
And it doesn't stop there.
Pursuing the logic of his method, Conway is able to create between in-between
numbers, then numbers between *these*, and so on, literally ad infinitum.
The result is limitless hierarchies of in-between numbers, never before
named in mathematics.
Conway's theory has ineffable graphic implications as well. Traditional
mathematical wisdom has it that a ruler's edge, a number line, is a blur
of points, each of which can be labeled with either a whole number, a
fraction, or an irrational number such as .1345792 ..., where the string
of digits goes on forever. All these points (or their numerical labels)
together are imagined to form a continuum, with no space between adjacent
points. Conway's theory, however, asks us to imagine numbers that fall
somehow between unimaginable cracks in this blur of points, and between
the cracks left behind by those numbers, and so on and so on. With his
theory, Conway has made credible what many persons before him had merely
speculated about: there is conceptually no limit to how many times an object
can be divided.
Conway's "All Numbers, Great and Small" shows off the boundless potential
of the null set, but also of the human mind. Human creative energy, like
nothing, isn't anything if it isn't potential. It is also an indomitable
part of being alive, as countless experiments have documented. People
who are deprived of their senses by being floated in silent, dark tanks
of water warmed to body temperature will hallucinate. It is as though
the human mind will not be stilled of its propensity to make something
of nothing even, or especially, when immersed in nothingness.
Like a physicist's vacuum, the human mind can be induced to create thoughts
that come seemingly out of nowhere. Mathematicians over the years have
documented this common phenomenon. The German Carl Friedrich Gauss recalled
that he had tried unsuccessfully for years to prove a particular theorem
in arithmetic, and then, after days of not thinking about the problem,
the solution came to him "like a sudden flash of lightning." The French
mathematician Henri Poincare, too, reported working futilely on a problem
for months. Then one day while conversing with a friend about a totally
unrelated subject, Poincare recalled that "... the idea came to me without
anything in my former thoughts seeming to have paved the way for it."
In this sense, the human mind is the real null set in Frege's and Conway's
number theories; the mathematical null set is but a subordinate entity
created after the mind's self-image."
PHYSICIST:
In the book "The Turning Point" Fritjof Capra (Ph.D in high-energy physics
from University of Vienna) writes (p.101):
"While the new physics was developing in the twentieth century, the
mechanistic Cartesian world view and the principles of Newtonian physics
maintained their strong influence on Western scientific thinking, and even
today many scientists still hold to the mechanistic paradigm, although
physicists themselves have gone beyond it.
...
In biology the Cartesian view of living organisms as machines, constructed
from separate parts, still provides the dominant conceptual framework.
Although Descartes' simple mechanistic biology could not be carried very
far and had to be modified considerably during the subsequent three hundred
years, the belief that all aspects of living organisms can be understood
by reducing them to their smallest constituents, and by studying the
mechanisms through which these interact, lies at the very basis of most
contemporary biological thinking. This passage from a current textbook
on modern biology is clear expression of the reductionist credo: 'One of
the acid tests of understanding an object is the ability to put it together
from its component parts. Ultimately, molecular biologists will attempt
to subject their understanding of cell structure and function to this sort
of test by trying to synthesize a cell.'
Although the reductionist approach has been extremely successful in biology,
culminating in the understanding of the chemical nature of genes, the basic
units of heredity, and in the unraveling of the genetic code, it nevertheless
has its severe limitations. As the eminent biologist Paul Weiss has
observed:
We can assert definitely ... on the basis of strictly empirical investiga-
tions, that the sheer reversal of our prior analytic dissection of the
universe by putting the pieces together again, whether in reality of
just in our minds, can yield no complete explanation of the behavior
of even the most elementary living system.
This is what most contemporary biologists find hard to admit. Carried
away by the success of the reductionist method, most notable recently in
the field of genetic engineering, they tend to believe that it is the only
valid approach, and they have organized biological research accordingly.
Students are not encouraged to develop integrative concepts, and research
institutions direct their funds almost exclusively to ward the solution
of problems formulated within the Cartesian framework. Biological phenomena
that cannot be explained in reductionist terms are deemed unworthy of
scientific investigation. Consequently biologists have developed very
curious ways of dealing with living organisms. As the distinguished biologist
and human ecologist Rene Dubos has pointed out, they usually feel most at
ease when the thing they are studying is no longer living.
...
An extreme case of integrative activity that has fascinated scientists
throughout the ages but has, so far, eluded all explanation is the phenome-
non of embryogenesis--the formation and development of the embryo--which
involves an orderly series of processes through which cells specialize
to form the different tissues and organs of the adult body. The interaction
of each cell with its environment is crucial to these processes, and the
whole phenomenon is a result of the integral coordinating activity of the
entire organism--a process far too complex to lend itself to reductionist
analysis. Thus embryogenesis is considered a highly interesting but quite
unrewarding topic for biological research.
...
Transcending the Cartesian model will amount to a major revolution in medical
science, and since current medical research is closely linked to research
in biology--both conceptually and in its organization--such a revolution
is bound to have a strong impact on the further development of biology."
***
I think it is quite interesting that "The Turning Point" was published
before "The Society of Mind" in reference to Fritjof Capra's comment,
"and even today many scientists still hold to the mechanistic paradigm."
Paradoxically, these three people's thoughts may sound unrelated. It is up
to you to decide, any comments?kevinc@auvax.UUCP (Kevin "auric" Crocker) (01/07/89)
In article <15152@mimsy.UUCP>, anderson@secd.cs.umd.edu (Gary Anderson) writes: >The successful application of artificial intelligence tools >and techniques will increase the degree to which humans depend >on machines. Clearly, human kind will be able to construct >smart, productive, efficient machines to carry out many important tasks >These machines will become superior to humans >along certain dimensions (ie computational speed, physical strength, agility, >stamina) in specific applications. As the performance of these machines >improves, and their use becomes more pervasive, we will come to >rely on them more and more for our survival and comfort. >These ubiquitous machines will touch the lives of nearly everyone and >they will be a source of great anxiety for many people. I couldn't help but have visions of Issac Asimov's robot and Foundation series flitter through my consciousness as I read the above. It sounds like such a nice productive arena to have all thoses robots doing all the tedious, meaningless, physically trying tasks that we humans really don't want to do. Hmm, doesn't sound a whole lot beyond the primative first stages of computers that we have now, what with Artificial Intelligence, and - I know the dirty words here- Expert systems. The almost endurable world that would permit humans to exercise their true `right' - devotion to thinking, and pleasure. Should I really include a |=#^ here or just a :-). Kevin Crocker -- Kevin "Auric" Crocker @Athabasca University {alberta ncc}auvax!kevinc
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (01/07/89)
From article <558@soleil.UUCP>, by peru@soleil.UUCP (Dave Peru): " ... " Paradoxically, these three people's thoughts may sound unrelated. It is up " to you to decide, any comments? Yes. Guillen (not Conway) doesn't make sense, and Minsky's and Capra's views seem to be in contradiction -- Minsky urging analysis into parts and Capra denigrating it. In the quoted passage from Capra, I believe one can detect some confusion among: (1) analysis into component parts (2) analysis into independently acting component parts (= Cartesianism?) (3) analysis in terms of more fundamental entities (= reductionism) It's hard for me to see that there can be any real objection to (1). I have been interested in analogues to the assumption of orthogonal axes, (2), for a long time, but have been unable to find any general discussions of the matter. Maybe someone can provide a reference? Here's a little example of this sort of reasoning from my own field. In 1783, in Elements of Phonetics, Geoffrey Holder pointed out that (voiceless) p, t, k are similar (voiced) to b, d, g, except for the action of the vocal cords, and that the latter are similar to (nasal) m, n, ng, except for the passage of air through the nose. He argued, on this basis, that there must exist in some language voiceless nasals -- this fills the gap in the paradigm. (It's very much like the prediction of new elements to fill holes in the periodic table.) Greg, lee@uhccux.uhcc.hawaii.edu
bwk@mbunix.mitre.org (Barry W. Kort) (01/07/89)
In article <552@soleil.UUCP> peru@soleil.UUCP (Dave Peru) asks: > Does anyone strongly disagree if I include in the definition of > "intelligence" the ability to recognize a paradox? Contributions to knowledge are made by people who recognize paradoxes in physics, math, philosophy, logic, etc. But their contributions come not from the act of recognition, but from acts of cognition which resolve the paradox. --Barry Kort
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (01/07/89)
In article <687@htsa.uucp> fransvo@htsa.UUCP (Frans van Otten) writes: >abilities essential for intelligence. These are mere examples. I really >stick to my definition of intelligence: > > *** Intelligence: The ability to draw a conclusion. > > *** Needed: A database and an algorithm to reach a conclusion > based on the data. > Wouldn't an if-then rule be intelligent under such a definition? I think you need more flexibility than that and prefer Dave's definition. There are many different aspects of intelligent. All intelligent entities, including humans, rarely can display all of the properties that people would like to say constitute intelligence. I suppose that is why the Turing test is so elegant.
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (01/08/89)
In article <558@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: > > Conway's "All Numbers, Great and Small" shows off the boundless potential > of the null set, but also of the human mind. Human creative energy, like > nothing, isn't anything if it isn't potential. It is also an indomitable > part of being alive, as countless experiments have documented. People > who are deprived of their senses by being floated in silent, dark tanks > of water warmed to body temperature will hallucinate. It is as though > the human mind will not be stilled of its propensity to make something > of nothing even, or especially, when immersed in nothingness. > This comment is quite naive, and I am surprise to find someone trained in physics can make it. Even though all sensory input is extinguished, the neural circuitry which is tuned to evaluate such input is still quite active. In the absence of input, it is the nature of such circuits to increase their sensitivity until artifactual output is obtained. The situation is somewhat analogous to amplifier circuits which go into oscillation when the gain is increased beyond a certain point. All neural tissues, including muscles do this. If you denervate a living muscle, it will, after a period of a few days, begin to twitch spontaneously. There is nothing mystical or infinite about such behavior. It can be explained on a purely mechanistic basis, and all neurologists are familiar with such behavior. >PHYSICIST: > >In the book "The Turning Point" Fritjof Capra (Ph.D in high-energy physics >from University of Vienna) writes (p.101): > > > We can assert definitely ... on the basis of strictly empirical investiga- > tions, that the sheer reversal of our prior analytic dissection of the > universe by putting the pieces together again, whether in reality of > just in our minds, can yield no complete explanation of the behavior > of even the most elementary living system. > All this says is that such systems, even the most elementary are very complex. As far as the nervous system is concerned, we are about at the level of snails (see the work of Eric Kandel, for example) in coming up with a more or less "complete" understanding of what is going on, at least on a macro level. Capra is definitely on the lunatic fringe on this subject. He embraces "holistic" medicine, chiropractic, and other even more bizarre medical quack systems which I suppose only enhance his popularity among his new age followers. He certainly isn't considered a touchstone among the physicists I know. I find little in his work to lead me to believe he knows anything substantial about the brain or biology.
bwk@mbunix.mitre.org (Barry W. Kort) (01/08/89)
In article <687@htsa.uucp> fransvo@htsa.UUCP (Frans van Otten) writes proposes a definition of intelligence: > *** Intelligence: The ability to draw a conclusion. > > *** Needed: A database and an algorithm to reach a conclusion > based on the data. > > *** Improvements: The ability to change the database. > The conclusion-algorithm being part of the database, > so that the system can add/change algorithms. > > I would like to know how other people think about my definition. I would suggest amending the first part to read "The ability to efficiently draw provably valid conclusions." This change suggests criteria for the improvements sought in the third part of your definition. Children (and adults) can easily draw invalid conclusions, but I'm not sure I want to label such behavior as highly intelligent. I am interested in the names of the distinct methods (algorithms) for efficiently drawing provably valid conclusions. So far, I have come up with: Adductive reasoning Deductive reasoning Inductive reasoning Inferential reasoning Model-based reasoning Combinatorial logic Intuitionist logic I would appreciate some discussion aimed at completing the list of known reasoning methods, together with their definitions. --Barry Kort
bwk@mbunix.mitre.org (Barry W. Kort) (01/08/89)
I continue to marvel at Dave Peru's fertile contributions to
our discussions in this newsgroup. The Minksy/Conway/Capra
excerpts were most stimulating.
Reductionist (analytical) reasoning is easy to describe and
easy to teach. But reductionism has a shortcoming.
If I give you a large, assembled jigsaw puzzle, and you examine
it piece by piece, you will end up with a pile of carefully
examined pieces. But you will have missed seeing the big
picture hidden in the assembled puzzle. This is called the
Forest and the Trees syndrome. After examining every element,
you must painstakingly reassemble them to see the big picture.
When you do so, you experience a profound psychological
transformation, called Insight or Epiphany. This rare and
treasured mental event is accompanied by a biochemical rush
of neurotransmitters such as serotonin, which yield a sense
of euphoria ("Eureka! I have found it!")
Another place reductionism fails is in the understanding of
emergent properties of circular systems. The simplest of
circular systems is the Furnace-and-Thermostat system.
When the furnace and thermostat are connected in a feedback
loop, the system exhibits the emergent property of maintaining
a stable room temperature in the face of unpredictable changes
in the outside weather. Feedback control theorists and
cyberneticians appreciate the emergent properties of circular
systems, but their appreciation is akin to seeing the big
picture in the pile of jigsaw puzzle pieces.
Minsky and Conway, and Gauss and Poincare engage in
synthetic reasoning (the complement of analytic reasoning).
Instead of understanding something by taking it apart,
they understand something by putting it together. It is
harder to teach synthetic reasoning. Artists and sculptors,
playwrights and poets, theoreticians and children -- these
are the synthetic thinkers, the practitioners of creative
intelligence.
The feedback loops of these discussion groups give rise to
an emergent property: the synthesis of ideas from diverse
quarters. The melting pot of ideas and the melding of minds
is the synthetic product of circular information channels.
--Barry Korthes@ecsvax.uncecs.edu (Henry Schaffer) (01/09/89)
In article <558@soleil.UUCP>, peru@soleil.UUCP (Dave Peru) writes: >... > MATHEMATICIAN/PHYSICIST/ASTRONOMY: > > In the book "Bridges To Infinity" Michael Guillen (Ph.D in physics, mathema- > matics, and astronomy from Cornell University) writes (p.98): ... > > Pursuing the logic of his method, Conway is able to create between in-between > numbers, then numbers between *these*, and so on, literally ad infinitum. > The result is limitless hierarchies of in-between numbers, never before > named in mathematics. Hmm, (even this has nothing to do with reductionism) how is this different that what is done in traditional mathematics? > ... > PHYSICIST: > > In the book "The Turning Point" Fritjof Capra [writes] ... > > Although the reductionist approach has been extremely successful in biology, > ... > As the eminent biologist Paul Weiss has observed: > > We can assert definitely ... on the basis of strictly empirical investiga- > tions, that the sheer reversal of our prior analytic dissection of the > universe by putting the pieces together again, whether in reality of > just in our minds, can yield no complete explanation of the behavior > of even the most elementary living system. > This seems to be an example of "proof by assertion". >... > An extreme case of integrative activity that has fascinated scientists ^^^^^^^^^^ - yes > throughout the ages but has, so far, eluded all explanation is the phenome- ^^^ - the large community of embryologists and developmental biologists would probably feel that they've explained *something* > non of embryogenesis-- ... > --a process far too complex to lend itself to reductionist analysis. ... Another proof by assertion - This whole controversy makes me think again about a question which has bothered me before. If reductionism is not sufficient - how can one show/prove that it is not sufficient. Clearly if a process is very complex, then much work must be done do reduce it sufficiently far to explain everything via a reductionist scenario. I doubt that any reductionist is willing to believe that embryogenesis is beyond reductionist analysis. We haven't even finished cataloging all the enzymes and enzymatic pathways in a single cell, there is still lots of (reductionist) work left to be done and which clearly can be done (I haven't heard anyone say that sequenceing the genome of a higher organism can't be done - just that it is a lot of work) and so clearly one can't give up on reductionism just yet. Is that the answer? One can't disprove reductionism as long as there is more work left to be done? That would mean essentially never. --henry schaffer n c state univ
bwk@mbunix.mitre.org (Barry W. Kort) (01/10/89)
In article <833@auvax.UUCP> kevinc@auvax.UUCP (Kevin "auric" Crocker)
refers to the "dirty word" of AI -- Expert Systems.
Although the expression hasn't caught on, I personally prefer
the phrase "Competent System" to "Expert System". It's not
as sexy sounding, but it may be a more accurate label.
--Barry Kortmarty@homxc.ATT.COM (M.B.BRILLIANT) (01/10/89)
In article <687@htsa.uucp>, fransvo@htsa.uucp (Frans van Otten) writes: > .... I really > stick to my definition of intelligence: > > *** Intelligence: The ability to draw a conclusion. > > *** Needed: A database and an algorithm to reach a conclusion > based on the data. > > *** Improvements: The ability to change the database. > The conclusion-algorithm being part of the database, > so that the system can add/change algorithms. I like this formulation because I think it contains the key to the definition of intelligence. An expert system can draw a conclusion using a database and an algorithm. I would not call it intelligent. Intelligence is a generalization of this capability, namely, the "Improvements" at the end of the list. It is not sufficient to be able to go one only step further, and have a database of databases, and a database of algorithms, and an algorithm for choosing new databases and new algorithms. Intelligence implies the ability to create an arbitrary number of levels of databases of databases, and databases of algorithms, and algorithms for choosing databases and algorithms from the preceding level. I suggest that the ability to reach an indefinite number of levels of generalization distinguishes intelligence from mere computation. M. B. Brilliant Marty AT&T-BL HO 3D-520 (201) 949-1858 Home (201) 946-8147 Holmdel, NJ 07733 att!houdi!marty1 Disclaimer: Opinions stated herein are mine unless and until my employer explicitly claims them; then I lose all rights to them.
EGNILGES@pucc.Princeton.EDU (Ed Nilges) (01/10/89)
In article <558@soleil.UUCP>, peru@soleil.UUCP (Dave Peru) writes: > > Then why do we so often embrace the strange idea that what we do is done > by Someone Else--that is, our Self? Because so much of what our minds > do is hidden from the parts of us that are involved with verbal > consciousness." > Interesting. Have you read Philosophy and the Mirror of Nature, by Richard Rorty, which deconstructs the notion of the self as the lonely spectator in an otherwise deserted movie theater (damned sad picture, not so?). According to Rorty, this notion got its start in the seventeenth century, and it is unnecessary. Edward Nilges "Where is the wisdom we lost in knowledge? Where is the knowledge we lost in information?" - T. S. Eliot
meadors@cogsci.ucsd.EDU (Tony Meadors) (01/10/89)
In article <558@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: >Please consider the following thoughts of three people concerning the physics >of the mind. > COMPUTER SCIENTIST: >In the book "The Society of Mind" Marvin Minsky writes (p.50): >"When people have no answers to important questions, they often give some > anyway. > What controls the brain? The Mind. > What controls the mind? The Self. > What controls the Self? Itself. > .... > It cannot help for you to think that > inside yourself lies someone else who does your work. This notion of > "hommunculus"--a little person inside each self--leads only to a paradox An infinite regress. One of the challenges of psychological explanation is to explain our overall intelligent behavior and cognitive abilities with a model whose parts are not themselves possessors of those abilities...this is how homunculi can creep into real world models. What Minsky is doing in the quoted passages is simply noting how commonsense notions such as self and mind entail the idea of a "detatched controller" and this quickly leads down the homunculi trail. > MATHEMATICIAN/PHYSICIST/ASTRONOMY: >In the book "Bridges To Infinity" Michael Guillen (Ph.D in physics, mathema- >matics, and astronomy from Cornell University) writes (p.98): > ........ > From there he goes on, however, to create an infinity of in-between numbers, > such as the number whose left set contains zero, {0}, and whose right set > contains one through infinity {1, 2, 3, ...}. > This defines a number somewhere > numbers, is embellished by an interminable number of in-between volumes. > And it doesn't stop there. > > Pursuing the logic of his method, Conway is able to create between in-between > numbers, then numbers between *these*, and so on, literally ad infinitum. > The result is limitless hierarchies of in-between numbers, never before > named in mathematics. I'm no mathematician, but if I take the numbers 2 & 3 and stick a bunch of new items between them (no matter how cleverly) I certainly won't have created "numbers never before named in mathematics." Numbers seem rather fixed to me, those that might be found on a simple numberline; the labels I attach to various points shouldn't make any difference...Unless these new numbers are not expressable in decimal form at all. If this is the case I missed the point but my point is below anyway... > points. Conway's theory, however, asks us to imagine numbers that fall > somehow between unimaginable cracks in this blur of points, and between > the cracks left behind by those numbers, and so on and so on. With his > theory, Conway has made credible what many persons before him had merely > speculated about: there is conceptually no limit to how many times an object > can be divided. Cosmic cracks eh. Again, Im not a numbers man, but was there ever any doubt that a given two points on a line one may always be found which lies between them? > Conway's "All Numbers, Great and Small" shows off the boundless potential > of the null set, but also of the human mind. Human creative energy, like > nothing, isn't anything if it isn't potential. It is also an indomitable > part of being alive, as countless experiments have documented. People > who are deprived of their senses by being floated in silent, dark tanks > of water warmed to body temperature will hallucinate. It is as though > the human mind will not be stilled of its propensity to make something > of nothing even, or especially, when immersed in nothingness. > > Like a physicist's vacuum, the human mind can be induced to create thoughts > that come seemingly out of nowhere. Mathematicians over the years have > documented this common phenomenon. The German Carl Friedrich Gauss recalled > that he had tried unsuccessfully for years to prove a particular theorem > in arithmetic, and then, after days of not thinking about the problem, > the solution came to him "like a sudden flash of lightning." The French > mathematician Henri Poincare, too, reported working futilely on a problem > for months. Then one day while conversing with a friend about a totally > unrelated subject, Poincare recalled that "... the idea came to me without > anything in my former thoughts seeming to have paved the way for it." > > In this sense, the human mind is the real null set in Frege's and Conway's > number theories; the mathematical null set is but a subordinate entity > created after the mind's self-image." I must say it's really getting deep at this point. I realize that the "wondrous parallels between profound mathematical principles with the human mind" is the idea here. But I see no more that a paper thin relatedness between the specifics under discussion. This reminds me of other cases where "deep fundamental" mathematic principles are put forward as "the essence" of thinking or mind (recursion a common one). Let's go over this again: > Conway's "All Numbers, Great and Small" shows off the boundless potential > of the null set, but also of the human mind. Human creative energy, like > nothing, isn't anything if it isn't potential. So roughly the claim is "the mind is like, the null set." (a california surfer dude accent would go nicely here). I find this a very strange claim but let's consider the two examples... First, > People > who are deprived of their senses by being floated in silent, dark tanks > of water warmed to body temperature will hallucinate. It is as though > the human mind will not be stilled of its propensity to make something > of nothing even, or especially, when immersed in nothingness. Yes people do eventually have all sorts of wild experiences. How does this relate to the mind being like a null set or the mathematical discussion at all? Does the null set notion PREDICT that those in such cahmbers will hallucinate? THERE IS ONLY A VERY CRUDE SEMANTIC RELATIONSHIP BETWEEN THE NULL SET AND SENSORY DEPRIVATION. "Oh, like both have to do with complete nothingness man..." Second, > Like a physicist's vacuum, the human mind can be induced to create thoughts > that come seemingly out of nowhere. Mathematicians over the years have > documented this common phenomenon. The German Carl Friedrich Gauss recalled >..... Yes, yes, such cases are well known. But now the relationship between the null set and the "example" is almost hard to find at all. First, there is no reason to suppose any sort of emptiness involved. Research on this "incubation" period of problem solving indicates that active though unconscious processing is involved in producting "the answer." And the individual, through his long and arduous pursuit of a solution to fulfill some set of constraints, has set up a situation where when the "answer" is unconsciously conceived of, it is "recognized" and brought to consciousness. Anyway THERE IS NOTHING MORE THAN A CRUDE SEMANTIC RELATIONSHIP BETWEEN THE NULL SET AND THE INCUBATION PHENOMENON IN PROBLEM SOLVING. > In this sense, the human mind is the real null set in Frege's and Conway's > number theories; the mathematical null set is but a subordinate entity > created after the mind's self-image." 1 THE HUMAN MIND IS NO MORE "THE REAL NULL SET IN...NUMBER THEORIES" THAN IT IS A BASEBALL BAT OR A TORNADO. 2 The notion that the null set arose as a mathematical concept due to man's perception of some nothingness within his psyche is absurd. > PHYSICIST: >In the book "The Turning Point" Fritjof Capra (Ph.D in high-energy physics >from University of Vienna) writes (p.101): > >"While the new physics was developing in the twentieth century, the > mechanistic Cartesian world view and the principles of Newtonian physics > maintained their strong influence on Western scientific thinking, and even > today many scientists still hold to the mechanistic paradigm, although > physicists themselves have gone beyond it. > ... > In biology the Cartesian view of living organisms as machines, constructed > from separate parts, still provides the dominant conceptual framework. > Although Descartes' simple mechanistic biology could not be carried very > far and had to be modified considerably during the subsequent three hundred > years, the belief that all aspects of living organisms can be understood > by reducing them to their smallest constituents, and by studying the > mechanisms through which these interact, lies at the very basis of most > contemporary biological thinking. So is this a tirade against a mechanistic approach, or the reductionist enterprise? They are not the same of course. >.... > Transcending the Cartesian model will amount to a major revolution in medical > science, and since current medical research is closely linked to research > in biology--both conceptually and in its organization--such a revolution > is bound to have a strong impact on the further development of biology." Yeah this sounds like Capra. I don't know what it would mean to "transcend the cartesian model", and no explanation of what that would be like is offered in this passage. If what is meant is to "look for causes and processes outside the normal realm of measurable cause and effect then I would say that its hogwash. If its just a childlike hope that taking new perspectives, sometimes a "systems" or "cybernetic" perspective may yield new insight into complex systems, then point taken. >Paradoxically, these three people's thoughts may sound unrelated. It is up >to you to decide, any comments? Yes, not only unrelated, they are unremarkable. Dave, your postings remain without peer in being provocative and interesting. But trust me, the "deep stuff" concerning minds and brains, the meta-psychology, is largely fluff. Move up the scientific foodchain a bit. You know the old saying, fact is stranger than fiction. Its never been more true than in psychology. Get down to real data and yet keep these larger questions in mind. Read about the bizzare dissociations brain damaged patients exhibit, study up on perceptual illusions, investigate the cases of extraordinary memories (people can literally tell you what shirt they wore or the change they made on a given day in 1966, and its not a trick or learned ability). Well, you get the picture...these sorts of phenomenon baffle and challenge, and if there are secrets to be found and profound changes to take place in how we understand the mind it will likely be fueled by these inexplicable sorts of data. tonyM
markh@csd4.milw.wisc.edu (Mark William Hopkins) (01/11/89)
In article <43472@linus.UUCP> bwk@mbunix (Barry Kort) writes: >Reductionist (analytical) reasoning is easy to describe and >easy to teach. But reductionism has a shortcoming. > >If I give you a large, assembled jigsaw puzzle, and you examine >it piece by piece, you will end up with a pile of carefully >examined pieces. I don't know about that. I solve most of my puzzles by classifying pieces on the basis of their shape and printed color, with little or no regard for the place where they fit in the "big" picture. Yet, I also claim that I'm solving the puzzle holistically in the process. The "big" picture always emerges out of the jumble of pieces near the end.
bwk@mbunix.mitre.org (Barry W. Kort) (01/11/89)
In article <6177@ecsvax.uncecs.edu> hes@ecsvax.uncecs.edu (Henry Schaffer) worries about the overthrow of reductionism: > This whole controversy makes me think again about a question which >has bothered me before. If reductionism is not sufficient - how can >one show/prove that it is not sufficient. Clearly if a process is >very complex, then much work must be done do reduce it sufficiently >far to explain everything via a reductionist scenario. > >... Clearly one can't give up on reductionism just yet. > > Is that the answer? One can't disprove reductionism as long as there >is more work left to be done? That would mean essentially never. I don't think anyone is suggesting that reductionism (or analysis of a complex system into its constituent elements) is a doomed activity. I think the argument is that additional insight is gained through synthetic reasoning (constructing novel systems from known pieceparts). Nature does this all the time. The cerebral cortex of the species Homo sapiens sapiens is believed to be one of the most complex systems found in nature. We learn by taking apart, and we learn by putting together. There is room (and need) for both activities. Personally, I find that, as a species, we devote more time to disassembly than to assembly, and I would like us to spend more time developing our creative intelligence. But I wouldn't want a world in which we have to choose between holism and reductionism. Both are essential ingredients in cognitive growth. Now and then, I even like to rest and simply enjoy what is. --Barry Kort
peru@soleil.UUCP (Dave Peru) (01/11/89)
>>In the book "The Society of Mind" Marvin Minsky writes (p.50): >>"When people have no answers to important questions, they often give some >> anyway. >> What controls the brain? The Mind. >> What controls the mind? The Self. >> What controls the Self? Itself. >> .... >> It cannot help for you to think that >> inside yourself lies someone else who does your work. This notion of >> "hommunculus"--a little person inside each self--leads only to a paradox In article <686@cogsci.ucsd.EDU> (Tony Meadors) writes: >An infinite regress. >One of the challenges of psychological explanation is >to explain our overall intelligent behavior and cognitive >abilities with a model whose parts are not themselves possessors of >those abilities...this is how homunculi can creep into real world models. > What Minsky is doing in the quoted passages is simply noting how >commonsense notions such as self and mind entail the idea of a "detatched >controller" and this quickly leads down the homunculi trail. I would like to humbly express my opinion about the way Marvin Minsky describes "hommunculus" as "leads only to paradox". Using the word "only" is misleading, like there's something wrong with hommunculus or even having a paradox. Or as you have stated, "simply noting how". Personally, these kind of statements in any explanation are not very satisfying, in fact, I start to get uncomfortable. All I'm saying, considering the subject matter, is simply that things never to turn out so simple. Or at least, seem so simple to me. "The idea of a single, central Self doesn't explain anything. This is because a thing with no parts provides nothing that we can use as pieces of explanation!" MM. If to explain something, you must have parts, then at some point you got to reduce down to physics. I think our knowledge in physics is great, but limited. Physicists might have egos as big as atomic blasts, but unfortunately God is still alive. This bothers me and is why I have problems with reductionist thinking. Einstein said God does not play dice, or was it God that said Einstein does not play dice. Anyway, as far as I know, according to our current knowledge of physics, God does play dice and is probably living in Atlantic City. Who knows, maybe Donald Trump is the second coming of Christ. :-) Seriously, is there anyone out there who really thinks reductionism can explain everything there is to be explain? >>In the book "Bridges To Infinity" Michael Guillen (Ph.D in physics, mathema- >>matics, and astronomy from Cornell University) writes (p.98): >> ........ >> From there he goes on, however, to create an infinity of in-between numbers, >> such as the number whose left set contains zero, {0}, and whose right set >> contains one through infinity {1, 2, 3, ...}. >> This defines a number somewhere >> numbers, is embellished by an interminable number of in-between volumes. >> And it doesn't stop there. >> >> Pursuing the logic of his method, Conway is able to create between in-between >> numbers, then numbers between *these*, and so on, literally ad infinitum. >> The result is limitless hierarchies of in-between numbers, never before >> named in mathematics. >I'm no mathematician, but if I take >the numbers 2 & 3 and stick a bunch of >new items between them (no matter how cleverly) >I certainly won't have created "numbers never >before named in mathematics." Numbers seem rather fixed to me, those that >might be found on a simple numberline; the labels I attach to various >points shouldn't make any difference...Unless these new numbers are not >expressable in decimal form at all. If this is the case I missed the >point but my point is below anyway... Don't waive this off, spend some time with this. What Conway does is really awesome. If fact, it defines the word awesome. The idea of "nothingness" as opposed to "nothing as something", i.e. the set {0}, is really neat! And then boom, all the rational and irrational numbers spring to life. To say "Numbers seem rather fixed to me" seems fixed or closed minded to me. >> points. Conway's theory, however, asks us to imagine numbers that fall >> somehow between unimaginable cracks in this blur of points, and between >> the cracks left behind by those numbers, and so on and so on. With his >> theory, Conway has made credible what many persons before him had merely >> speculated about: there is conceptually no limit to how many times an object >> can be divided. > >Cosmic cracks eh. >Again, Im not a numbers man, but was there ever any doubt that a given two >points on a line one may always be found which lies between them? "Cosmic", interesting word choice. When you were younger, did you ever get the feeling while you were half asleep that you were falling off your bed? You suddenly wake up as you slam your hand down on the mattress. I have this feeling all the time, but nothing to slam against. :-) And mathematically speaking, the way Conway generates numbers is the closest thing I've seen to expressing this feeling. >> People >> who are deprived of their senses by being floated in silent, dark tanks >> of water warmed to body temperature will hallucinate. It is as though >> the human mind will not be stilled of its propensity to make something >> of nothing even, or especially, when immersed in nothingness. > >Yes people do eventually have all sorts of wild experiences. How does this >relate to the mind being like a null set or the mathematical discussion at >all? I knew I should have left the float-tank part out. People have all kinds of prejudices. Tony, have you ever floated? I haven't, but maybe Guillen has. Apparently, Guillen thought the experience related to the discussion to use the analogy. You think the analogy doesn't apply, okay. I still think it's a neat idea and I'll reserve judgement until after I've floated. > Does the null set notion PREDICT that those in such cahmbers will >hallucinate? THERE IS ONLY A VERY CRUDE SEMANTIC RELATIONSHIP BETWEEN >THE NULL SET AND SENSORY DEPRIVATION. "Oh, like both have to do with >complete nothingness man..." This California surfer stuff is indicative of your close mindedness and adds nothing to the conversation. Which is appropriate considering the subject matter. When you say "there is only a very crude semantic relationship between the null set and sensory deprivation" are you speaking from experience? >> In this sense, the human mind is the real null set in Frege's and Conway's >> number theories; the mathematical null set is but a subordinate entity >> created after the mind's self-image." > >1 THE HUMAN MIND IS NO MORE "THE REAL NULL SET IN...NUMBER THEORIES" > THAN IT IS A BASEBALL BAT OR A TORNADO. > >2 The notion that the null set arose as a mathematical concept due to > man's perception of some nothingness within his psyche is absurd. Considering the quality of your comments, your mind is a perfect example of the null set. All you've really said is this is bullshit with bullshit reasons. Maybe this is all we can ever say about this subject. If you see something that is blatently wrong then say so and state why. However, if these interpretations are simply contrary to your own interpretations or intuition, then don't come off so condescending with words like "absurd". Like you know better, maybe you do. Personally, my belief system is evolving. I remain open to new ideas. >> Transcending the Cartesian model will amount to a major revolution in medical >> science, and since current medical research is closely linked to research >> in biology--both conceptually and in its organization--such a revolution >> is bound to have a strong impact on the further development of biology." > >Yeah this sounds like Capra. I don't know what it would mean to "transcend >the cartesian model", and no explanation of what that would be like is >offered in this passage. If what is meant is to "look for causes and >processes outside the normal realm of measurable cause and effect >then I would say that its hogwash. I think what Capra means by "transcend the cartesian model" is that a human being as an organism is affected by the environment in such a way that some processes will not be explanable out of that context. Things may be so interconnected that reductionism may be inadequate. I think this is interesting when you consider the relationship of the mind in respect to understanding the physics of the environment or the physics of the mind. > If its just a childlike hope that >taking new perspectives, sometimes a "systems" or "cybernetic" >perspective may yield new insight into complex systems, then >point taken. Childlike? I don't understand. What distinguishes childlike from adultlike? >>Paradoxically, these three people's thoughts may sound unrelated. It is up >>to you to decide, any comments? > > Yes, not only unrelated, they are unremarkable. Then why did you make a remark. I was trying to show some ideas about and of the mind in respect to the reductionist approach. Some people liked it. > Dave, your postings remain without peer in being provocative and > interesting. But trust me, the > "deep stuff" concerning minds and brains, the meta-psychology, > is largely fluff. Trust you? Is it safe? :-) Some fluff hardens. I think alot of people have been a little hard on Guillen. This guy has some really neat things to say. Consider from his essay "Irrational Thinking" from his book "Bridges to Infinity" (p.38-39): "Despite this preeminence of rational numbers, science does need irrational numbers. For well over a century, scientists have been taking note of a growing inventory of special quantities whose appearance in nearly every scientific theory signifies their import in the modern description of space-time. These natural constants can be seen as nature's vital statistics, and right now it looks as though every one of them is an irrational number. For example, one of these constants, the speed of light, has been measured out to nine decimal places, and the digits have yet to show any pattern. (Expressed in millions of meters per second, our best measurement of the speed of light is the number .299792458.) Another constant is one that is descriptive of dynamic behavior at the atomic level. It is called the fine-structure constant, and there is no pattern to its digits even when measured out to ten decimal places. (Our best measurement of the fine- structure constant, which is a dimensionless quantity, is .0072973502.) In physics alone there are more than a dozen of these constants, which have been measured out to anywhere from a few to eleven decimal places, and not one of them has a pattern to its digits." When I read this I was astonished. Of course, some of these constants may not be irrational numbers. But what would be really awesome is to come up with some physics that would predict these irrational numbers. Anyway, some more fluff for the pile. > Move up the scientific foodchain a bit. You know > the old saying, fact is stranger than fiction. Its never been more true > than in psychology. Get down to real data and yet > keep these larger questions in mind. Read about the bizzare > dissociations brain damaged patients exhibit, study up on perceptual > illusions, investigate the cases of extraordinary memories (people can > literally tell you what shirt they wore or the change they made on > a given day in 1966, and its not a trick or learned ability). Well, > you get the picture...these sorts of phenomenon baffle > and challenge, and if there are secrets to be found and profound changes > to take place in how we understand the mind it will likely be fueled > by these inexplicable sorts of data. I try to get down to real data as much as I can. That's why I like USENET, after I read all the fluff, I can see what real people think. In reference to "move up the scientific foodchain", I'm currently reading Paul Kennedy's "Rise and Fall of Great Powers". I want to find out why it is so hard nowadays for a person my age to buy a house.
dmocsny@uceng.UC.EDU (daniel mocsny) (01/11/89)
In article <331@csd4.milw.wisc.edu>, markh@csd4.milw.wisc.edu (Mark William Hopkins) writes: > In article <43472@linus.UUCP> bwk@mbunix (Barry Kort) writes: > >If I give you a large, assembled jigsaw puzzle, and you examine > >it piece by piece, you will end up with a pile of carefully > >examined pieces. > I don't know about that. I solve most of my puzzles by classifying pieces > on the basis of their shape and printed color, with little or no regard > for the place where they fit in the "big" picture. > > Yet, I also claim that I'm solving the puzzle holistically in the process. > The "big" picture always emerges out of the jumble of pieces near the end. How about those irritating puzzles with large washes of featureless background (sky, ocean, forest). Even with our terrific holistic pattern-matching power, the best we can often do is try every combination of pieces to see which ones fit together (the problem gets worse when a malicious jigsaw operator makes similar cuts that permit close but erroneous fits). Assembling a solid-color puzzle reduces us to the level of a slow, awkward serial computer, with perhaps some advantage in avoiding certain obvious misfits. Is the solid-color puzzle problem NP-complete? Then again, I don't know anyone who has spent enough time assembling solid-color puzzles to perform at an expert level. Perhaps subtle cues exist that would allow our holistic power to get its foot in the door and fetch the complexity monster a swift kick. A portion of the puzzle with more information content provides suitable grist for our holistic mill. Fixing the position of a piece is "only" a matter of spotting when the pattern on it corresponds uniquely to a detail on the puzzle box (if the puzzle came in a plain box, toss it in the incinerator), or to a partially-assembled structure. The holistic pattern-matcher must work in the face of rotations, and know when to ignore or exploit the shape of a particular puzzle piece. But I think I am subverting Barry's original comment. He seemed to be saying that the way the puzzle happens to divide into pieces has _nothing_at_all_to_do_ with the picture that appears on the puzzle. The "obvious" reductionist approach to "understanding" or "explaining" the picture on the puzzle is doomed from the start. However, I think the futility of the reductionist approach here follows from the nature of the puzzle. I.e., the puzzle is an artifact, and as such its decomposition is arbitrary. Do we in fact see such "arbitrary" or misleading decomposition in nature, or do we explain our failure to explain as due to nothing more than our limited knowledge and bookkeeping ability? Cheers, Dan Mocsny dmocsny@uceng.uc.edu "God is subtle, but He is not malicious." -- A. Einstein "Stop telling God what to do." -- P.A.M. Dirac (?) (Sorry, science historians, if I botched this one)
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (01/11/89)
In article <564@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: >God is still alive. This bothers me and is why I have problems with >reductionist thinking. Einstein said God does not play dice, or was it God >that said Einstein does not play dice. Einstein did say "God does not play dice with the universe", and one of his friends (I think it was Pauli) finally retorted: "When are you going to quit telling God what to do?" > >Seriously, is there anyone out there who really thinks reductionism can explain >everything there is to be explain? > I doubt if the human race will survive long enough to explain everything there is to explain whatever method is used. That isn't the point. The point is, when dealing with complex systems, reductionism is a necessary step if we are to understand them. Only a first step, since then we have to learn how to assemble the reduced parts back into a whole again. But it has worked splendidly in the past and there is no sign at all that it is exhausted as a method, despite the ravings of Capra and others. This all has nothing whatever to do with God. If reductionism allows us to make progress in understanding all parts of the universe we have heretofore investigated, why should the same method not work in the case of the human mind?
mark@verdix.com (Mark Lundquist) (01/12/89)
In article <687@htsa.uucp> fransvo@htsa.UUCP (Frans van Otten) writes: >I really stick to my definition of intelligence: > > *** Intelligence: The ability to draw a conclusion. > > *** Needed: A database and an algorithm to reach a conclusion > based on the data. > > *** Improvements: The ability to change the database. > The conclusion-algorithm being part of the database, > so that the system can add/change algorithms. > >I would like to know how other people think about my definition. Well...I'm not sure that it's very useful. The problem is that it doesn't really answer anything for us; instead, it takes all the interesting and difficult issues that used to be part of the question "What is intelligence?" and pushes them back into different questions. For instance, "How do we describe the particular unique way that human beings draw conclusions?" This would be perfectly valid if this definition accurately reflected what the word 'intelligence' does in our language. Unfortunately, it doesn't (I think), unless you mean something very special by the phrase "draw a conclusion". Suppose I said "Well, I have this tic-tac-toe-playing program. It must be intelligent, since it correctly draws a conclusion about what the next move should be." One might respond by saying, "Not really; your program can't be said to hold a _belief_ about the correctness of the proposed tic-tac-toe move. You've simply set up the bits in the machine in such a way that it prints something on the screen, which happens to correspond to a correct tic-tac-toe move, but only because you've rigged it that way. It's a conclusion in a sense, but it's not the kind of conclusion that I meant". This of course would beg the question "Then just what sort of 'conclusion' do you mean?" However, one might respond to my tic-tac-toe suggestion as follows: "You're quite right. Your tic-tac-toe program _is_ intelligent. Of course, it's far less intelligent than a baboon. Humans, in turn, exhibit intelligence on a grander scale yet. But in principle, it's the same." This response would also be question-begging. How is it that humans and baboons apply this principle, to be able to exhibit their respective degrees of intelligence? No matter how you slice it, it comes up peanuts. I would say that intelligence is what minds do. Of course, this definition is _at_least_ as question-begging as the last one, and almost as useless, except for one thing: it does seem to describe what people mean when they use the word 'intelligence'. I suspect that we'll never find a definition of intelligence that escapes the difficulty that I've described. I guess I still don't understand the necessity of formulating such a definition.
stevev@uoregon.uoregon.edu (Steve VanDevender) (01/12/89)
In article <564@soleil.UUCP> peru@soleil.UUCP (Dave Peru) writes: >I think alot of people have been a little hard on Guillen. This guy has >some really neat things to say. Consider from his essay "Irrational Thinking" >from his book "Bridges to Infinity" (p.38-39): > >"Despite this preeminence of rational numbers, science does need irrational > numbers. For well over a century, scientists have been taking note of a > growing inventory of special quantities whose appearance in nearly every > scientific theory signifies their import in the modern description of > space-time. These natural constants can be seen as nature's vital statistics, > and right now it looks as though every one of them is an irrational number. > For example, one of these constants, the speed of light, has been measured > out to nine decimal places, and the digits have yet to show any pattern. > (Expressed in millions of meters per second, our best measurement of the > speed of light is the number .299792458.) Another constant is one that is > descriptive of dynamic behavior at the atomic level. It is called the > fine-structure constant, and there is no pattern to its digits even when > measured out to ten decimal places. (Our best measurement of the fine- > structure constant, which is a dimensionless quantity, is .0072973502.) In > physics alone there are more than a dozen of these constants, which have > been measured out to anywhere from a few to eleven decimal places, and not > one of them has a pattern to its digits." > >When I read this I was astonished. Of course, some of these constants may >not be irrational numbers. But what would be really awesome is to come up >with some physics that would predict these irrational numbers. > >Anyway, some more fluff for the pile. > This is definitely fluff. It is ridiculous to try to read meaning into the digits of the number for the speed of light in meters per second. Meters and seconds are entirely arbitrary measurements of space and time and it's not surprising that physical constants are going to show no patterns when expressed in base ten and when measured in metric units. You should know that in many physics problems, measurements are normalized so that, say, c is 1. The values of the constants themselves are not important. The relationships between the values of physical constants are. It is slightly more interesting to consider dimensionless constants like the fine structure constant, which are independent of any measuring system. However, who is to say that there is no pattern to its digits when we only have ten decimal places and uncertainty to work with, and we're looking at it in base ten anyway? When the Planck length is on the order of 10^-23 meters, is ten or eleven decimal digits of a constant enough to make a pronouncement on? Guillen's title "Irrational Thinking" may apply to more than just his essay. To claim that numbers we can only measure to finite precision and which involve uncertainty are therefore irrational is, well, irrational. Rational numbers are perfectly adequate for representing the physical constants he talks about. From what I've seen of Guillen so far, I can see why people are hard on him. He may have fascinating mystic insights but his attempts to justify them in scientific or mathematical terms don't work. The best I can say about his attempt to make an analogy between creating a continuum of numbers out of the null set and the ability of the mind to produce unpredictable thoughts is that the analogy is strained. Does he show that the mind produces some insights out of nothing? No. Can he know that it does? I think not. It is just as tenable to say that a mind produces insights via processes that are not accessible to that mind's own consciousness, from information it already has. This also counters the justification that sensory deprivation somehow shows that the mind makes something out of nothing. People who climb into a tank have memories, and when they start to hallucinate their minds presumably aren't creating visions out of nothing--they're creating hallucinations out of what is already in their minds. Would a mind that is completely blank, with no prior experiences, and that is deprived of all input hallucinate? Is this experiment possible? Probably not. Guillen isn't talking about this experiment, anyway, but it's what he really should be talking about if he wants to claim that a mind can generate something from nothing like Conway's theory can generate numbers from the null set. I think the reductionism/holism argument boils down to what I think is a pair of clearer questions: Is the universe explainable by rules? Can those rules be derived by observing the universe? Science assumes that the answer to both of those questions is "yes." My understanding of holism leads me to think that it would answer "no" to one or both of those questions. -- Steve VanDevender stevev@drizzle.cs.uoregon.edu "Bipedalism--an unrecognized disease affecting over 99% of the population. Symptoms include lack of traffic sense, slow rate of travel, and the classic, easily recognized behavior known as walking."
bwk@mbunix.mitre.org (Barry W. Kort) (01/13/89)
In article <331@csd4.milw.wisc.edu> markh@csd4.milw.wisc.edu (Mark William Hopkins) takes me up on my jigsaw puzzle metaphor: >In article <43472@linus.UUCP> bwk@mbunix (Barry Kort) writes: >>If I give you a large, assembled jigsaw puzzle, and you examine >>it piece by piece, you will end up with a pile of carefully >>examined pieces. > >I don't know about that. I solve most of my puzzles by classifying pieces >on the basis of their shape and printed color, with little or no regard >for the place where they fit in the "big" picture. > >Yet, I also claim that I'm solving the puzzle holistically in the process. >The "big" picture always emerges out of the jumble of pieces near the end. I grant that you are solving the puzzle holistically. After all, the big picture does in fact emerge at the end. But the *process* of solution seems to be occuring outside the focus of concious attention. We can teach people how to examine the jigsaw pieces, and classify them by color, shape, and texture. But the method of assembly which yields the "Aha! Insight" seems to a fuzzier, less algorithmic activity. Perhaps it is occuring largely in the right hemisphere, using parallel processing and combinatorial logic. Why is it that holistic thinking and insight seems to come during periods of sleep or during periods when our attention is diverted away from the problem at hand? Why is it that the solution "shows up" without warning? --Barry Kort
bwk@mbunix.mitre.org (Barry W. Kort) (01/13/89)
In article <564@soleil.UUCP> peru@soleil.UUCP (Dave Peru) opines: > To say "Numbers seem rather fixed to me" > seems fixed or closed minded to me. In Howard Rheingold's book, _Tools of Thought_, there is a sketch of the neurophysiologist and pioneering cyberneticist, Warren McCulloch. As Rheingold repeats the story, McCulloch was an abnormally gifted and colorful person who had a firm background in mathematics. A teacher asked McCulloch what he wanted to do with his obviously brilliant future. "Warren," said he, "what is thee going to be?" And I said, "I don't know," "And what is thee going to do?" And again I said, "I have no idea, but there is one question I would like to answer: What is a number, that man may know it, and a man that he may know a number?" He smiled and said, "Friend, thee will be busy as long as thee lives." > What distinguishes childlike from adultlike? On weekends I work as a volunteer in the Children's Discovery Room at the Boston Museum of Science. Occasionally I ask a parent, "What is the difference between a child and a scientist?" Most of them quickly respond, "No difference?" I often feel sorry for adults who have lost their childlike curiousity somewhere along the way. Fortunately a few children grow up to be scientists. It is a shame that so many people become adulturated enroute to maturity. --Barry Kort Today's Quote: "Nothing is as simple as it seems at first, or as hopeless as it seems in the middle, or as finished as it seems in the end."
bwk@mbunix.mitre.org (Barry W. Kort) (01/13/89)
In article <568@uceng.UC.EDU> dmocsny@uceng.UC.EDU (Daniel Mocsny) writes: > Is the solid-color puzzle problem NP-complete? There are two kinds of extra-hard jigsaw puzzles: the solid-color puzzles (Little Red Riding Hood's Hood) and the puzzle in which all the pieces are the same shape (Schmuzzles). But curiously enough, the solid-color Schmuzzle puzzle isn't even NP-hard. It's NP-ridiculous. :-) On a more sublime note, Dan returns to the original point of discussion: > But I think I am subverting Barry's original comment. He seemed to > be saying that the way the puzzle happens to divide into pieces > has _nothing_at_all_to_do_ with the picture that appears on the > puzzle. The "obvious" reductionist approach to "understanding" or > "explaining" the picture on the puzzle is doomed from the start. I suppose Mother Nature is not so devilish as the jigsaw puzzle maker. But our own category boundaries are still somewhat arbitrary. And, by studying the "elements" we don't automatically understand how they assemble themselves into "molecules". What I am saying is that anaylysis and differentiation are valuable tools, but creative intelligence also requires synthesis and integration. --Barry Kort
mark@verdix.com (Mark Lundquist) (01/14/89)
In article <686@cogsci.ucsd.EDU> meadors@cogsci.UUCP (Tony Meadors) writes: > "deep stuff" concerning minds and brains, the meta-psychology, > is largely fluff. Move up the scientific foodchain a bit. You know > the old saying, fact is stranger than fiction. Its never been more true > than in psychology. Get down to real data and yet > keep these larger questions in mind. Read about the bizzare > dissociations brain damaged patients exhibit, study up on perceptual > illusions, investigate the cases of extraordinary memories (people can > literally tell you what shirt they wore or the change they made on > a given day in 1966, and its not a trick or learned ability). Well, > you get the picture...these sorts of phenomenon baffle > and challenge, and if there are secrets to be found and profound changes > to take place in how we understand the mind it will likely be fueled > by these inexplicable sorts of data. Try any of the books written by Oliver Sacks ("A Leg To Stand On", "The Man Who Mistook His Wife For A Hat", etc). These books are accounts of some really strange disorders experienced by patients who had had trauma to the right hemisphere of the brain. These disorders profoundly change the patients' whole experience of being a human being. Their symptoms are not easily measured or quantified, and the disorders (according to Sacks) do not lend themselves well traditional case studies. Sacks decided that the appropriate form of 'case study' for these disorders is the story. He tells these stories with acumen, compassion, insight, and humor. He's also got another book (I can't remember the title) in which he discusses the relationships between Parkinson's and Tourette's syndromes.
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (01/15/89)
From article <43582@linus.UUCP>, by bwk@mbunix.mitre.org (Barry W. Kort): " ... But our own category boundaries are still somewhat arbitrary. " And, by studying the "elements" we don't automatically understand how " they assemble themselves into "molecules". Of course not, but is this a fair analogy for reductionism? I don't think so. Reductionist theories may occasionally arise by identifying elements apart from the patterns they assemble into (perhaps molecular biology would be a case?), but more typically the pattern is observed first. Later, a reduction into elements which assemble according to certain rules is proposed to explain the patterns. There is no step of analysis apart from synthesis -- the rules of assembly are intrinsically a part of the theory. Instances are the analysis of atoms to explain the pattern of the periodic table and the analysis of particles to explain the 8-fold way, probably. An instance I know more about may be drawn from Ventris' decipherment of Linear B. The molecules were, let us say, the signs of the writing system, and the atoms the vowels and consonants they stand for. A pattern in the data was that some signs appeared only at the beginning of (what were inferentially) words. One basis of the decipherment was the identification of such signs as standing for single vowels, the reasoning being that if the script was syllabic and if the language being written did not permit vowels to occur next to one another within a word (which is a common restriction), vowel-only syllables and their signs could occur only word-initially. This would explain the pattern. This, and other such inferences comprised Ventris' theory. (Other previous failed attempts to decipher the script were, however, based on direct assignments of phonetic values to signs.) One cannot find here a step of synthesis that is chronologically or logically "after" the analysis. I suspect that the criticism proponents of holistic theories make of reductionism is founded on a play on words -- an equivocation. There is traditionally a use of the term 'analysis' which opposes it to synthesis, but more commonly, 'analysis' does not refer merely to a decomposition somehow apart from composition. Greg, lee@uhccux.uhcc.hawaii.edu
mirk@warwick.UUCP (Mike Taylor) (01/16/89)
In article <1995@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >If reductionism allows us to make progress in understanding all parts >of the universe we have heretofore investigated, why should the same >method not work in the case of the human mind? Because the human mind is, by its very nature, something that can only be observed in its entirety from within, and this viewpoint of conciousness that we have is not succeptible to reductionist methods because we cannot view the phenomenon objectively. It is an intrinsically subjective thing. Thomas Nagel's article "What is it like to be a bat?" (which can be found in Hofstadter & Dennet's "The Mind's I") makes this point in rather more detail, but in a very dull and dry way, IMHBDO. His basic point is that we cannot understand what it is like to be a bat because it is a feeling subjective to the bat (if it is conscious at all). We can imagine what it would be like for ourselves to be a bat - but to have a true picture of the phenomenon of bat-consciousness, we must understand what it is like for the bat to be bat. Clear? No, I didn't think so :-( :-) I will try to restate the point in its bare form: to analyse something by reductive techniques, we must be able to view it objectively. But to view consciousness objectively is to omit the most important aspect of the phenomenon, namely the subjective experience of it, and thus any reductionist anaysis made on this basis will be incomplete and/or inaccurate. There - that wasn't so bad, was it? :-) ______________________________________________________________________________ Mike Taylor - {Christ,M{athemat,us}ic}ian ... Email to: mirk@uk.ac.warwick.cs *** Unkle Mirk sez: "Em9 A7 Em9 A7 Em9 A7 Em9 A7 Cmaj7 Bm7 Am7 G Gdim7 Am" *** ------------------------------------------------------------------------------
bwk@mbunix.mitre.org (Barry W. Kort) (01/17/89)
In article <3498@uoregon.uoregon.edu> stevev@drizzle.UUCP (Steve VanDevender) writes: > I think the reductionism/holism argument boils down to what I think is > a pair of clearer questions: Is the universe explainable by rules? > Can those rules be derived by observing the universe? Science assumes that > the answer to both of those questions is "yes." My understanding of > holism leads me to think that it would answer "no" to one or both of those > questions. Steve, did you really mean "derive" rather than "discover"? Einstein supposed that the universe would appear the same to all observers. From this supposition, he derived the Theory of Relativity. His starting point was neither a discovery nor a derivation. But he discovered that his derivation led to predictions which were borne out by experimental observation. But Einstein's nemesis was the Quantum Theory with it's dice-playing lack of rhyme or reason. So one of the "rules" appears to be lawless and chaotic behavior. Whether Stephen Hawking and others will ultimately imagine/discover/derive rules underlying quantom randomness remains to be seen. Personally, I believe that quantum indeterminacy will survive the razor of Occam, and that we will end up thanking our "lucky stars" for the gift of life, including intelligent life. --Barry Kort
abrown@homxc.ATT.COM (A.BROWN) (01/17/89)
Can someone please E-mail the difference between reductionism and
non-reductionism. I'm doing a paper on Artificial Intelligence.
Which argues that given the proper sample space, computers can
adequately simulate the 'Primitive Visual System'. I now need to
conclude but am stuck as to whether this validates either system.
Thanks a million
abrownlee@uhccux.uhcc.hawaii.edu (Greg Lee) (01/18/89)
From article <5038@homxc.ATT.COM>, by abrown@homxc.ATT.COM (A.BROWN): " " Can someone please E-mail the difference between reductionism and " non-reductionism... The Encyclopedia of Philosophy has some stuff under Laws and Theories, Reductionism, which begins: "Since theories do not refer directly to observables, at least prima iacie, and do not make directly testable statements, the first attempt to clarify their status was the suggestion that they make a disguised reference to observables; that is, that they provide some kind of shorthand for observation statements, or that their content can be exhaustively translated into or reduced to observation statements. ..." The article opposes reductionism to instrumentalism and realism. So far as I can tell, this "proper" sense of 'reductionism' has no relation to the way the term was being used in the recent discussion in these groups, where it meant 'science'. Greg, lee@uhccux.uhcc.hawaii.edu
throopw@xyzzy.UUCP (Wayne A. Throop) (01/18/89)
> mark@verdix.com (Mark Lundquist) > [...] one might respond to my tic-tac-toe suggestion as follows: > "You're quite right. Your tic-tac-toe program _is_ intelligent. Of course, > it's far less intelligent than a baboon. Humans, in turn, exhibit > intelligence on a grander scale yet. But in principle, it's the same." > This response would also be question-begging. How is it that humans and > baboons apply this principle, to be able to exhibit their respective degrees > of intelligence? This response does not beg the question at all. Or rather, it has a simple and straightforward answer for the question Mark claims it begs. The tic-tac-toe program models a game. The human models the game, the world, the human's self, the relationship among these entities, and on and on. The baboon (presumably) has fewer models of less accuracy than does the human. Or to put it another way, the answer to the question Mark poses is, humans and baboons apply the same principles as does the tic-tac-toe program, but they apply them to more things, more richly, and more accurately. Before anybody thinks I'm saying that AI is a SMOP, (to add lots of models and make them rich and accurate) let me assure you all that I *don't* minimize the difficulties or the unknowns in this way. After all, it is not known how one goes about building rich and accurate models of things, and tying these to perceptions. All I'm saying is that the position of "understanding is modeling" is not an obviously flawed position to take, nor does the position lack something as obvious as a distinguishing factor between levels of understanding. -- God made integers, all else is the work of man. --- Leopold Kronecker -- Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw
demers@beowulf.ucsd.edu (David E Demers) (01/18/89)
In article <2894@xyzzy.UUCP> throopw@xyzzy.UUCP (Wayne A. Throop) writes: > >Before anybody thinks I'm saying that AI is a SMOP, ^^^^ Um, are we all supposed to know what this means? >-- >Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw ^^^^^ a maze of twisty mailpaths, all different... Dave DeMers demers@cs.ucsd.edu
litow@csd4.milw.wisc.edu (Bruce E Litow) (01/18/89)
Recently some postings have appeared in which the type of argument (so-called) indicated in the summary has been invoked to maintain that reductionist methods cannot succeed in mind studies. I cannot accept that we can use the construction: `the mind by its very nature ...' when we haven't a clue as to what that `very nature' might be. In arguments based on this construction one is always forced at some point into actually accepting that there is a mind in toto which escapes whatever approach is being argued against. That is the mind is an entity. (Following Rilke perhaps the ``Angels'' see it entire) Once this is admitted ,then mind study is on par with physics which also faces a unity (the universe) about which all our understanding has come from reductionist methods. An interesting extended attempt in support of the claim that mind studies cannot proceed via reduction is given in Fodor's ``Modularity of Mind''. However,Fodor only makes the case for cognition being beyond our present reductions and nothing more. I believe that there is tremendous confusion in mind studies between e.g. general,metaphysical speculation about mind and reductions such as neurophysiology,molecular physiology,linguistic inquiries,etc. The first is limited because it only rarely can provide testable hypotheses. It is unscientific. Its utility comes from inspiring people to examine things but it is useless for carrying out research about mind. We have nothing else but reduction when it comes to science.
bettingr@sunybcs.uucp (Keith E. Bettinger) (01/19/89)
In article <906@ubu.warwick.UUCP> mirk@uk.ac.warwick.cs (Mike Taylor) writes: > > [ ... ] > >I will try to restate the point in its bare form: to analyse something >by reductive techniques, we must be able to view it objectively. But >to view consciousness objectively is to omit the most important aspect >of the phenomenon, namely the subjective experience of it, and thus >any reductionist anaysis made on this basis will be incomplete and/or >inaccurate. > >There - that wasn't so bad, was it? :-) Maybe it IS so bad. Let me try taking this position to its logical conclusion. We study things objectively, but consciousness is "subjectivity" itself, so that any objective study of subjectivity will be just missing the essence of what is being studied. Thus, we should reject objectivity for the one case of studying subjectivity. But we have already rejected subjective studies of phenomena, because its person-dependent results are only useful to the person making the discovery. Eliminating both means of investigation, this argument would seem to obviate any reductive attempt to fully study consciousness. Does this, then, argue for holism? I'm sure this line of reasoning has a gaping hole in it, but it seems interesting on the surface. (Another in a series of Why-bother? arguments.) Pardon my laziness, but I leave it to the esteemed members of the net to plunge the dagger into the heart of this demon... ------------------------------------------------------------------------- Keith E. Bettinger "Paradise SUNY at Buffalo Computer Science Is exactly like Where you are right now CSNET: bettingr@Buffalo.CSNET Only much much BITNET: bettingr@sunybcs.BITNET Better" - Laurie Anderson INTERNET: bettingr@cs.buffalo.edu UUCP: ..{bbncca,decvax,dual,rocksvax,watmath,sbcs}!sunybcs!bettingr -------------------------------------------------------------------------
geb@cadre.dsl.PITTSBURGH.EDU (Gordon E. Banks) (01/19/89)
In article <906@ubu.warwick.UUCP> mirk@uk.ac.warwick.cs (Mike Taylor) writes: >In article <1995@cadre.dsl.PITTSBURGH.EDU> geb@cadre.dsl.pittsburgh.edu (Gordon E. Banks) writes: >>If reductionism allows us to make progress in understanding all parts >>of the universe we have heretofore investigated, why should the same >>method not work in the case of the human mind? > >Because the human mind is, by its very nature, something that can only >be observed in its entirety from within, and this viewpoint of conciousness >that we have is not succeptible to reductionist methods because we cannot >view the phenomenon objectively. It is an intrinsically subjective thing. > Certainly the mind can not be observed in its entirety from within. Introspection is a very poor tool for understanding the mind. If we were able to understand the hardware (wetware) in which the mind is implemented, and create simulations which show similar behavior to minds, then don't you think we would be able to better understand the natural mind? Especially since we could perform experiments with the simulations which we cannot do easily with the mind?
bwk@mbunix.mitre.org (Barry W. Kort) (01/20/89)
In article <2894@xyzzy.UUCP> throopw@xyzzy.UUCP (Wayne A. Throop) writes: > After all, it is not known how one goes about building rich and > accurate models of things, and tying these to perceptions. All > I'm saying is that the position of "understanding is modeling" is > not an obviously flawed position to take, nor does the position > lack something as obvious as a distinguishing factor between > levels of understanding. As it happens, I build models for a living. And, Wayne's comment notwithstanding, I think I know how I do it. :-) In my work, I like to think of the modeling step as "capturing the structure and behavior of the real system with the model." Note that the English word, "comprehension", means "to capture with." So I agree with Wayne that modeling is a way of understanding (comprehending) something. After I capture/comprehend/understand the system, I use the model to think about what would happen if I tweak one of the "knobs" on the model. I like to think of this activity as "cognition". I can then add automatic feedback loops which maintain stable behavior under nominal perturbing influences. Next, I like to subject the model to abnormal conditions (like overloading it, or introducing a fault condition). Then I can observe the behavior and see how the feedback loops compensate for my nefarious disruption and disturbance of the peace. Since this step provides awareness of cause-and-effect patterns, I call this step "consciousness". Finally, I collect all the observations of behavior under overload and fault conditions, and learn how to map the observable symptoms back to the organic cause. The diagnostic model is the inverse model of the original system. So I call this step "insight." The whole process is referred to as CCCI (C-cubed I): Comprehension, Cognition, Consciousness, and Insight. --Barry Kort
throopw@xyzzy.UUCP (Wayne A. Throop) (01/25/89)
> bwk@mbunix.mitre.org (Barry W. Kort) >> throopw@xyzzy.UUCP (Wayne A. Throop) >> After all, it is not known how one goes about building rich and >> accurate models of things, and tying these to perceptions. > As it happens, I build models for a living. And, Wayne's comment > notwithstanding, I think I know how I do it. :-) My statement above is intended to mean that we don't know of any algorithms that can build models as rich and accurate as those we humans build automatically or at least "subconsciously" (at least, not for as broad a domain and with reasonable efficency). In particular, the step in model building algorithm that Barry calls "comprehension" or "capture" is not itself an algorithm, or an atomic operation of any automaton known. It is in that narrow sense that I meant that "it is knot known how one goes about building rich and accurate models of things". -- English language human-machine dialog is roughly equivalent to a conversation between an idiot and a very young child. --- Vincent Rauzino - Datamation May 1982 -- Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw
bwk@mbunix.mitre.org (Barry W. Kort) (01/28/89)
In article <3046@xyzzy.UUCP> throopw@xyzzy.UUCP (Wayne A. Throop) discusses the non-algorithmic aspect of model construction. Wayne writes: > the step in model building algorithm that Barry calls "comprehension" > or "capture" is not itself an algorithm, or an atomic operation of any > automaton known. The selection of the substrate upon which to build a model is indeed a non-systematic step. It takes an Eastman, or a Land to find the apporpriate substrate. But after that, the transcription from real object to image can be utterly mechanical, as we see in Xerography and photography. On Saturdays, I work as a volunteer in the Children's Discovery room at the Boston Museum of Science. One of the educational toys found in the room is a construction set called Marble Works. The plastic pieces can be put together to build wonderful structures through which a marble rolls from top to bottom. One day I discovered that I could construct a tower in the form of a double helix. Two marbles raced down the intertwined paths and came out at separate exits. A few weeks later, I coached a 9-year old boy to build a double helix as his bemused father looked on. It wasn't until later that it dawned on him that the boy had constructed a model of DNA. The lad was just mechanically assembling the parts according to a suggested pattern. --Barry Kort