throopw@agarn.dg.com (Wayne A. Throop) (03/14/89)
> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) > I am a > holist, but I don't see how an attribute of a part can be transferred > to the whole if it doesn't exist in the part. No problem, the systems reply doesn't claim that understanding was "transfered" to the whole system from Searle... it claims that the whole system understands as an emergent property, as a consequence of the arrangement of the rules for which Searle just happens to be an interpreter. Understanding is a property of a process interacting (however indirectly) with the object of understanding. > The interesting thing > about systems is the attributes of the whole which CANNOT be attributes > of the parts, not true here I'm afraid. Why isn't it true here? I'm under the impression that it IS true here. -- "This is just the sort of thing that people never believe." --- Baron Munchausen -- Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (03/16/89)
In article <4079@xyzzy.UUCP> throopw@agarn.dg.com (Wayne A. Throop) writes: >> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) >> about systems is the attributes of the whole which CANNOT be attributes >> of the parts, not true here I'm afraid. > >Why isn't it true here? I'm under the impression that it IS true here. I mustn't be clear. Usually, a system possesses attributes which no part *CAN* possess, and thus does not possess. Here, the part Searle can possess understanding. The issue is one of attributes common to a/some part(s) and the emergeant system. I don't know an example where a system has attributes that a part CAN have, but does not have. Equilibrium, for example, can never be an attribute of a part (unless it is a system). As Searle is not a system, this doesn't apply. -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert
geddis@polya.Stanford.EDU (Donald F. Geddis) (03/18/89)
In article <2599@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: >Usually, a system possesses attributes which no part *CAN* possess, >and thus does not possess. > >Here, the part Searle can possess understanding. > >The issue is one of attributes common to a/some part(s) and the >emergeant system. I don't know an example where a system has >attributes that a part CAN have, but does not have. Counterexample: The program MACSYMA running on a particular computer has the attribute CAN-SOLVE-SYMBOLIC-CALCULUS-EQUATIONS (CSSCE attribute). Now, give me a copy of the source code for MACSYMA and I'll hand-simulate it. Now the system (Me + Source Code) can solve all sorts of complex equations that I alone can't. However, I do have the capacity to have the CSSCE attribute, I just don't happen to have it at this moment. So there's an example where a system has attributes that a part can have, but does not have. -- Don -- Geddis@Polya.Stanford.Edu "Ticking away the moments that make up a dull day..." -- Pink Floyd
harnad@elbereth.rutgers.edu (Stevan Harnad) (03/18/89)
SYMBOL LOGIC 101 (MAKEUP)
geddis@polya.Stanford.EDU (Donald F. Geddis) of
Organization: Stanford University
" The program MACSYMA running on a particular computer has the attribute
" CAN-SOLVE-SYMBOLIC-CALCULUS-EQUATIONS (CSSCE attribute). Now, give me a
" copy of the source code for MACSYMA and I'll hand-simulate it.
"
" Now the system (Me + Source Code) can solve all sorts of complex equations
" that I alone can't. However, I do have the capacity to have the CSSCE
" attribute, I just don't happen to have it at this moment.
"
" So there's an example where a system has attributes that a part can have,
" but does not have.
I'll rephrase it cryptically, since it's all been said longhand, in vain,
so many times before:
(1) It is not in dispute that systems can have attributes that their parts
do not have. What is in dispute is what systems, what parts, what attributes.
(2) It is not in dispute that Searle has the capacity to understand
Chinese. He just does not happen to understand it at the moment.
(3) There is no basis whatever (I HOPE everyone agrees) for projecting
Searle's undisputed actual capacity for understanding English now, and
potential for understanding Chinese in the future, onto anything at
all, part or whole. I hope everyone sees THAT's just double-talk...
(4) The attribute of being able to solve equations is not the same as the
attribute of understanding.
(5) Neither is the attribute of being able to manipulate Chinese
symbols (even under the counterfactual hypothesis that one can
manipulate them well enough to pass the LTT) the same as the attribute
of being able to understand Chinese symbols. Why (for those who thought
it might have been)? One reason is Searle's Chinese Room Argument.
(Here and in my papers I've given several others, including the
"symbol-grounding problem.")
(6) What is the simple conclusion of this simple argument that someone
who has understood it must draw -- unless he has a valid
counterargument (or has become unalterably soft-wired to the
simple-minded belief that thinking is just symbol crunching)? That
thinking is not just symbol crunching. Q.E.D. (R.I.P.)
Refs: Searle, J. (1980) Minds, Brains and Programs. Behavioral and Brain
Sciences 3: 417-457
Harnad, S. (1989) Minds, Machines and Searle. Journal of Experimental
and Theoretical Artificial Intelligence 1: 5 - 25.
--
Stevan Harnad INTERNET: harnad@confidence.princeton.edu harnad@princeton.edu
srh@flash.bellcore.com harnad@elbereth.rutgers.edu harnad@princeton.uucp
BITNET: harnad@pucc.bitnet CSNET: harnad%princeton.edu@relay.cs.net
(609)-921-7771throopw@agarn.dg.com (Wayne A. Throop) (03/21/89)
> harnad@elbereth.rutgers.edu (Stevan Harnad) > [..it is not the case that..] > the attribute of being able to manipulate Chinese > symbols (even under the counterfactual hypothesis that one can > manipulate them well enough to pass the LTT) the same as the attribute > of being able to understand Chinese symbols. Why (for those who thought > it might have been)? One reason is Searle's Chinese Room Argument. More like "Chinese Room Anecdote". Given that its sole force of argument is gained by assuming its concluson by appeal to anthropomorphism, there are many people who don't find it convincing. Not that I think any near descendant of "Doctor" or other toys should be thought of as posessing understanding. Just that Searle has failed to prove (or even show reason) that a much more distant and complicated descendant should not be thought of in this way. > (Here and in my papers I've given several others, including the > "symbol-grounding problem.") And since the symbols of a symbol crunching system are every bit as well grounded as those humans use (at least potentially), this reason is somewhat less than convincing also. ( Note: I don't contend that they are as well grounded by the criteria Steve Harnad would like to use. I just think that some irrelevant criteria are snuck in there. ) > [...] thinking is not just symbol crunching. Q.E.D. (R.I.P.) And claiming that these anecdotes constitute "proof" of anything at all, especially in the course of pursuing "argument by insult and intimidation" combined with "argument by emphatic assertion" is (need I say it?) equally unconvincing. -- "Who would be fighting with the weather like this?" "Only a lunatic." "So you think D'Artagnian is involved?" --- Porthos, Athos, and Aramis. -- Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw
throopw@agarn.dg.com (Wayne A. Throop) (03/21/89)
>,>>> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) >> throopw@dg-rtp.dg.com (Wayne Throop) >>> about systems is the attributes of the whole which CANNOT be attributes >>> of the parts, not true here I'm afraid. >>Why isn't it true here? I'm under the impression that it IS true here. > Usually, a system possesses attributes which no part *CAN* possess, > and thus does not possess. > Here, the part Searle can possess understanding. But the relevant part, the one pointed to to (attempt to) show the shortcomings of the CR, is not "Searle", but rather "Searle blindly following rules". "Searle blindly following rules" cannot (by definition of what it means to blindly follow rules) understand. Even if Searle goes out and learns Chinese, "Searle blindly following rules" does not, in any relevant sense, understand Chinese. The systems reply is that Searle-plus-the-rules *does* understand. ( Of course, Searle's reply to the systems reply is moot, because he assumes the conscious mind of Searle has access to all capabilities of all systems operating in the physical system of Searle's brain, despite the copious evidence that this is not the case (eg, savant talents, multiple personalities). ) In fact, the whole CR argument is a simple and blatant appeal to anthropomorphism, an anecdote with no force of formal reasoning behind it. It doesn't "prove" anything at all, and convinces only those who already agree with its disguised anthropocentric premises. -- "Who would be fighting with the weather like this?" "Only a lunatic." "So you think D'Artagnian is involved?" --- Porthos, Athos, and Aramis. -- Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw
ellis@unix.SRI.COM (Michael Ellis) (03/22/89)
> Wayne A. Throop >The systems reply is that Searle-plus-the-rules *does* understand. That requires a leap of faith I'm willing to make for other humans, animals (and martians should they arrive), but not for an artifact whose design fails to include whatever relevant causes there may be to consciousness itself. >In fact, the whole CR argument is a simple and blatant appeal to >anthropomorphism, an anecdote with no force of formal reasoning behind >it. It doesn't "prove" anything at all, and convinces only those >who already agree with its disguised anthropocentric premises. What "anthropocentric" premises might those be? That subjective experience and intentional states with semantic content are real? That the subject under study is the human mind? -michael
fransvo@htsa.uucp (Frans van Otten) (03/23/89)
Michael Ellis writes: >Wayne A. Throop writes: > >>The systems reply is that Searle-plus-the-rules *does* understand. > >That requires a leap of faith I'm willing to make for other >humans, animals (and martians should they arrive), but not >for an artifact whose design fails to include whatever >relevant causes there may be to consciousness itself. The word "understanding" has three very different meanings. I think the main problem in this entire discussion is that most people keep mixing those meanings. *** First interpretation of "understanding" I can say "I understand". This statement is based on some feeling; it is subjective. This might be represented/implemented by a flag which is set to "true" or "false", or by a variable which contains a value within a (continuous) range, where there is some critical value when I get the feeling that I do understand. *** Second interpretation of "understanding" I can also say "You understand". This refers to some other entity then me myself. This is the "understanding" which is "measured" by the Turing Test (whatever variant you are using). Then first of all, there is the "other minds problem". Does the entity I am referring to have a mind ? I don't know. But I do say that it understands. So I might conclude: External behaviour resulting in me stating "you understand" does NOT require that the entity has a mind. And there was the argument "when you know the used algorithm, you usually don't say that it understands". I disagree on this one. Let's take a calculator. I can say "it understands how to add two numbers". Let's take an email-program. When I mail to someone on my local computer, the program "understands" not to call some other system for that message. When I mail to some unknown system, it does understand that the message must be sent to some backbone. My point is that in the process of thinking about things like "understanding", "consciousness", "intelligence" etc. many people lose their normal interpretations of those words and start to require undefined things, just because "human understanding is so very special, how could some simple device like a computer ever be able to have it ?". *** Third interpretation of "understanding" Neither of the above described interpretations of "understanding" mentions the actual process involved in understanding. There is no hint as to how understanding might work, what it might be. The interpretation of "understanding" concerns this process or state, the actual implementation of understanding a concept. The Chinese Room argument mixes the three described meanings of "understanding" in such a way that nobody knows what is true and what is false: 1. We have a set of rules [a computer program] defining which Chinese characters should be presented in response to incoming Chinese characters. It is assumed that when someone strictly follows those rules, the other person(s) involved in this conversation could be fooled into believing they actually are having a correspondence with a human being (or some other entity) understanding what is written. This concerns the second interpretation of understanding I described: "you understand": the Turing Test is passed. 2. Then the argument continues: "the person interpreting the rules doesn't understand Chinese". This concerns the first interpretation of understanding (the subjective one). 3. Then finally, the conclusion is made: "as this person, who is doing everything a computer would do, does not understand Chinese, then a computer can't understand Chinese either". This refers to the third interpretation of understanding, the objective one, the kind AI-researchers are (or should be) interested in. Objections against this argument have been: 1. The set of rules would be far too big, too complex to be executed by a human in a reasonable amount of time, etc., the argument relies on intuitions which are misguided thus. Of course, the set of rules would be highly complex. But the Chinese Room argument was a thought experiment, so that is no problem (but it helps the confusion). Would it have been possible for Einstein to travel on a lightbeam ? 2. The Turing Test tests on bevioural characteristics of the system, not on the internal (cognitive) functions. Very true, but that was exactly why the Turing Test was used in the argument. The idea was "make it seem to understand, this can be measured by the Turing Test, then see inside to find out whether or not it does understand". 3. The person executing the rules is not able to feel the understanding of the system. That is true, too. This is one of the flaws of the argument. But it doesn't touch the heart of it. We are not looking for a system that has a feeling that it does understand. We are trying to find a system that (objectively) understands. My objection against the Chinese Room argument is that the different interpretations of the word "understanding" are mixed in a way that is not tolerable. First of all, behavioural characteristics tests as the Turing Test cannot produce any evidence. The best they can do is "it seems to understand" or "it doesn't seem to understand". Drew Mcdermott was very clear on this is his paper he posted on the net. Secondly, as many posters wrote, you can't base a conclusion about an entire system on observations (let alone feelings) of a single part of it. This has been shown in many ways. The best way was probably "let someone execute the rules of physics of his own body; although the person understands things, he wouldn't understand those things from calculating his own physics". Note: The question has been raised, if the person would understand the rules, would he then understand Chinese ? I think Stevan Harnad is right on saying "no"; the symbol grounding problem does exist ! When you know that you should respond with a certain Chinese character when you see another certain one, this doesn't make you understand Chinese ! My theory on this is, that understanding requires representation of the external symbols (be it Chinese characters, vocal representations of words, red traffic lights, or whatever) in internal symbols. You can base actions on your internal symbols, but not on external symbols. Then the "denotation/connotation" argument of Roel Wieringa shows up. How do I translate external symbols to internal symbols ? The process of reaching a conclusion based on an internal state (I call this process "intelligence") is independent of the external symbols and their meanings. But the rules that are used to reach a conclusion are based on the socially accepted meanings of the external symbols. The same holds for the translation from external to internal symbols. This is what Karl Kluge meant when he wrote about communicating his desire for a bowl of icecream to his appartmentmate. Final note: If you ever manage to make a system that does understand, pack it in a green humanoid container, put this in a rocket, write in big letters "From Mars" on it, and launch this rocket to Michael Ellis. -- Frans van Otten Algemene Hogeschool Amsterdam Technische en Maritieme Faculteit fransvo@htsa.uucp
throopw@agarn.dg.com (Wayne A. Throop) (03/25/89)
> ellis@unix.SRI.COM (Michael Ellis) > [..the "systems reply" claim that the system posesses the understanding..] > requires a leap of faith I'm willing to make for other > humans, animals (and martians should they arrive), but not > for an artifact whose design fails to include whatever > relevant causes there may be to consciousness itself. I'm not willing to make it easy either. The commotion is over just what constitutes "relevant causes". I simply do not think that the human brain has any mysterious "causal powers" that a computer executing a suitable program does not. >> [..the CR argument..] >> doesn't "prove" anything at all, and convinces only those >> who already agree with its disguised anthropocentric premises. > What "anthropocentric" premises might those be? That subjective > experience and intentional states with semantic content are real? > That the subject under study is the human mind? No. The appeal is made through the argument that since Searle doesn't understand, and since Searle (the human component) is where we would normally look for understanding in such a system, the system must not understand. The conclusion simply doesn't follow. It's rather like finding a corpse stabbed to death, showing that the butler didn't do it, and concluding that there was no homicide. -- If someone tells me I'm not really conscious, I don't marvel about how clever he is to have figured that out... I say he's crazy. --- Searle (paraphrased) from an episode of PBS's "The Mind" -- Wayne Throop <the-known-world>!mcnc!rti!xyzzy!throopw
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (03/28/89)
In article <4506@xyzzy.UUCP> throopw@agarn.dg.com (Wayne A. Throop) writes: >I'm not willing to make it easy either. The commotion is over just >what constitutes "relevant causes". I simply do not think that the >human brain has any mysterious "causal powers" that a computer >executing a suitable program does not. OK then, let's here what a "suitable" program would be. I contend that AI research doesn't have a grasp of what "suitable" means at all. For one, human minds are not artefacts, whereas computer programs always will be. This alone will ALWAYS result in performance differences. Given a well-understood task, computer programs will out-perform humans. Given a poorly understood task, they will look almost as silly as the author of the abortive program. The issue as ever is what we do and do not understand about the human mind, the epistemelogical constraints on this knowledge, and the ability of AI research as it is practised to add anything at all to this knowledge. Come on then boys and girls in AI, lets hear it on "suitable" :-) Cyberpunk fragments are acceptable, and indeed indistinguishable in a LTT from some AI research :-] -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert
maddoxt@novavax.UUCP (Thomas Maddox) (03/29/89)
In article <2599@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: >I mustn't be clear. His motto. Tom Maddox UUCP: ...{ucf-cs|gatech!uflorida}!novavax!maddoxt
fransvo@htsa.uucp (Frans van Otten) (03/30/89)
Gilbert Cockton writes: >In article <4506@xyzzy.UUCP> throopw@agarn.dg.com (Wayne A. Throop) writes: >>I'm not willing to make it easy either. The commotion is over just >>what constitutes "relevant causes". I simply do not think that the >>human brain has any mysterious "causal powers" that a computer >>executing a suitable program does not. [...] >For one, human minds are not artefacts, whereas computer programs >always will be. This alone will ALWAYS result in performance >differences. Given a well-understood task, computer programs will >out-perform humans. Given a poorly understood task, they will look >almost as silly as the author of the abortive program. This discussion has degraded into a fight between two groups with different viewpoints: 1. Humans have some mysterious powers that are responsible for their having a mind. Animals might also have these powers, maybe even martians. This property might be inherent to the building material; carbon-hydrogen has it, Si doesn't. 2. Understanding etc. are properties which arise from a certain way to process information. The information theory is what matters, not the way it is implemented. If we humans can do it using our hardware (neurons etc), then computers are able to do this using theirs. I believe that those who support 1. are in an ideological grippe. This is an unsupported way of looking at things. If these people might think they could find support in religions, I have to dissapoint them. In no religion known to me it is stated that the mind/spirit/... (the non-physical thing) is dependent (in its being) on its body (the physical thing). This includes religions ranging from Christianity to Bhuddism, Zen and Sufi. There is not either any support for this viewpoint from the technological world. There is no apparent chemical reason why carbon-hydrogene molecular groups can and Si-molecules can not give rise to something as high-level as understanding and consciousness. So I think these people are stuck somewhere between a "rational" and a "not-rational" (emotional/...) viewpoint, but are too lazy to really think about the issue. When they join a discussion, it becomes a mess. My personal opinion on this is as follows. In the evolutionary process, with "survival of the fittest", you have to behave in such a way that you will survive long enough to raise a new generation. As the level of complexity of the organism increases, it will have to do more "information processing": to find food, to protect against enemies, etc. My point: intelligence etc. developed out of a need to determine how to behave in order to survive. So the behaviourist approach is justified: "when the system seems to act intelligently, it *is* intelligent". Then we invented the computer. We start wondering: can we make this machine intelligent ? Before we can write a program for this, we must understand the algorithm humans use. This proves to be very difficult. Research is hindered by people claiming that understanding requires very mysterious causal powers which computers, due to their design, can never have. Gilbert Cockton even claims that because human minds are not artifacts, while computer systems always will be, there will always be performance differences. Apart from the fact that this statement is nonsense, it is not of any importance to AI-research. -- Frans van Otten Algemene Hogeschool Amsterdam Technische en Maritieme Faculteit fransvo@htsa.uucp
jb@aablue.UUCP (John B Scalia) (03/30/89)
In article <2691@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: > [Some deletions here] >For one, human minds are not artefacts, whereas computer programs >always will be. This alone will ALWAYS result in performance >differences. Given a well-understood task, computer programs will >out-perform humans. Given a poorly understood task, they will look >almost as silly as the author of the abortive program. Right, Gilbert. I could argue equally well that anything/person performing a poorly understood task will always be outperformed by a person/thing accomplishing a well understood task. This has to be true even if both "laborors" are human. Else, we'd all have children born who can already speak all languages as well as performing integral calculas, instead of adults who despite years of "programming" cannot read or write. Hey, the "wiring" is in place, but it takes years of "programming" to make it accomplish much. On this same vein, computer programs are still much simpler running on much more simple hardware, but they are still deeply related and much more guaranteable. I believe here lies most of the AI problem: most of us have a hard time with a computer program which produces an error where a human being erring is not only possible but often expected. -- A A Blueprint Co., Inc. - Akron, Ohio +1 216 794-8803 voice UUCP: {uunet!}aablue!jb Marriage is a wonderful institution, but who FidoNet: 1:157/697 wants to spend their life in an institution. EchoNet: US:OH/AKR.0
jackson@freyja.css.gov (Jerry Jackson) (03/31/89)
In article <813@htsa.uucp>, fransvo@htsa (Frans van Otten) writes: >Gilbert Cockton writes: >This discussion has degraded into a fight between two groups with different >viewpoints: > > 1. Humans have some mysterious powers that are responsible for their > having a mind. Animals might also have these powers, maybe even > martians. This property might be inherent to the building material; > carbon-hydrogen has it, Si doesn't. This is a seriously flawed statement of the position. It is not that carbon "has something" that silicon doesn't -- that would be *stupid*. What is claimed is that possibly it is not merely functional structure that determines the mind. The "silicon-based" computers we have are brain-like only in functional organization (if that :-). Perhaps consciousness is a *chemical* phenomenon and not a result of a particular functional structure. If a computer could be built based on carbon instead of silicon, the argument would be the same. > > 2. Understanding etc. are properties which arise from a certain way to > process information. The information theory is what matters, not > the way it is implemented. If we humans can do it using our hardware > (neurons etc), then computers are able to do this using theirs. > >I believe that those who support 1. are in an ideological grippe. This >is an unsupported way of looking at things. If these people might think >they could find support in religions, I have to dissapoint them. In no >religion known to me it is stated that the mind/spirit/... (the non-physical >thing) is dependent (in its being) on its body (the physical thing). This >includes religions ranging from Christianity to Bhuddism, Zen and Sufi. If anyone in this group were to appeal to religions for support they might as well put "Don't pay attention to this article!" in the subject line. Pointing out that no major religion supports a point of view is irrelevant. > >There is not either any support for this viewpoint from the technological >world. There is no apparent chemical reason why carbon-hydrogene molecular >groups can and Si-molecules can not give rise to something as high-level as >understanding and consciousness. > See above. >So I think these people are stuck somewhere between a "rational" and a >"not-rational" (emotional/...) viewpoint, but are too lazy to really think >about the issue. When they join a discussion, it becomes a mess. No comment. > >My personal opinion on this is as follows. In the evolutionary process, >with "survival of the fittest", you have to behave in such a way that you >will survive long enough to raise a new generation. As the level of >complexity of the organism increases, it will have to do more "information >processing": to find food, to protect against enemies, etc. My point: >intelligence etc. developed out of a need to determine how to behave in >order to survive. So the behaviourist approach is justified: "when the >system seems to act intelligently, it *is* intelligent". > I think most people involved in this argument assume that humans evolved to their present state. This, however, is beside the point. Yes, if one wishes to define intelligence "from the outside", it is perfectly ok to do so. Searle and others are simply arguing that something is left out when one does so. This may not make a practical difference in system performance at all. The main point of the CR thought experiment is that there is a subjective experience that is usually labeled "understanding" that would appear to be missing from the CR. This doesn't mean the room can't behave identically to a human speaker of chinese. In fact, the very argument presupposes that it *can*. Humans have a strange attribute known as subjectivity that doesn't immediately appear to be reducible to structure or functional organization. It may even be totally unnecessary for intelligent behavior. If so, though, it is hard to imagine why such a thing would evolve. Some people seem to misunderstand why "pain", for instance, is considered to be problematic for machine intelligence. A common point of view I have seen on the net goes something like this: The computer has sensors that determine when it is damaged or likely to be damaged. These send a signal to the central processor which takes appropriate action.. (like saying "Ouch!" :-). This hardly explains pain! The signal in question fulfills the same functional role as a signal in the human nervous system.. i.e. indicating a hazard to the body. The only thing missing is the *pain*! To use an example I have used before, ask yourself why you take aspirin for a headache. I claim it is not because you contemplate the fact that a signal is travelling through your body and you wish it would stop. You take the aspirin because your head *hurts*. The functionalist model would map a pain signal to some quantity stored in memory somewhere... Does it really make sense to imagine: X := 402; -- OW! OW! 402, ohmigod!... X := 120; WHEW!.. thanks! I can imagine a system outputting this text when the quantity X changes, but I can't honestly imagine it actually being in pain.. Can you? >Then we invented the computer. We start wondering: can we make this >machine intelligent ? Before we can write a program for this, we must >understand the algorithm humans use. This proves to be very difficult. >Research is hindered by people claiming that understanding requires very >mysterious causal powers which computers, due to their design, can never >have. Gilbert Cockton even claims that because human minds are not >artifacts, while computer systems always will be, there will always be >performance differences. Apart from the fact that this statement is >nonsense, it is not of any importance to AI-research. >-- > Frans van Otten > Algemene Hogeschool Amsterdam > Technische en Maritieme Faculteit > fransvo@htsa.uucp --Jerry Jackson
pluto@beowulf.ucsd.edu (Mark E. P. Plutowski) (03/31/89)
In article <448@esosun.UUCP> jackson@freyja.css.gov (Jerry Jackson) writes: >... What is claimed is that >possibly it is not merely functional structure that determines >the mind. The "silicon-based" computers we have are brain-like only in >functional organization (if that :-). Perhaps consciousness is a *chemical* >phenomenon and not a result of a particular functional structure. Perhaps this is true, at some level of analysis. But it is surely false, at some level of implementation. If consciousness includes all mental calculations, then witness the computation facilitated by topological mappings, interconnection schemata, and by the compartmentalization of these "chemical phenomenon." > [such a human trait] as subjectivity ... doesn't immediately appear to be reducible to structure >or functional organization. It may even be totally unnecessary for intelligent >behavior. If so, though, it is hard to imagine why such a thing would evolve. > Depending upon your perspective, perhaps. But, it should not be difficult to imagine *how* such a thing would evolve. How can we not be subjective, unless out of body experience is a fact? Indeed, it is more likely that subjectivity is a result of the underlying implementation of intelligence, rather than a refuting piece of evidence against the possibility that intelligence depends upon its underlying structure. If, for all practical purposes, we can divorce consciousness from its underlying structure, then, why is one particular practical purpose so difficult for individuals to achieve? Namely, objectivity. Mark Plutowski =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- You can skin a gift horse in the mouth, but you can't make him drink. ---------------------------------------------------------------------- Mark Plutowski INTERNET: pluto%cs@ucsd.edu Department of Computer Science, C-014 pluto@beowulf.ucsd.edu University of California, San Diego BITNET: pluto@ucsd.bitnet La Jolla, California 92093 UNIX:{...}ucsd!cs.UCSD.EDU!pluto =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- You can skin a gift horse in the mouth, but you can't make him drink,
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (03/31/89)
>This discussion has degraded into a fight between two groups with different >viewpoints: There is considerable diversity, as well as incompatability, in the arguments both for and against the possibility of strong AI. You are particularly poor in your grasp of all the anti-AI arguments. Some are based on the impossibility of simulating the brain's hardware on a digital computer (indeed, the impossibility of accurately and faithfully simulating ANY part of the natural world on a computer). Others rely on epistemic arguments. Others rely on theories of ideology which deny any possible objective status to such value laden concepts as 'intelligence', which are symptomatic of a system of social stratification peculiar to modern Europe (and taken on in an even cruder form in the New World). The word doesn't have a usable meaning in any scientific context. The sensible approach is to identify tasks for automation, describe them accurately and acceptably, and then proceed. Designing systems to possess ill-defined and hardly understood properties is an act of intellectual dishonesty at worst, and an act of intellectual mediocrity at best. Robotics has the advantage of dealing with fairly well-defined and understood tasks. I'd be surprised if anyone in robotics really cares if their robots are 'intelligent'. Succesful performance at a task is what matters. This is NOT THE SAME as intelligent behaviour, as we can have clear conditions for success for a task, but not for intelligent behaviour. Without verification or falisification criteria, the activity is just a load of mucking about - a headless chicken paradigm of enlightenment. >with "survival of the fittest", you have to behave in such a way that you >will survive long enough to raise a new generation. As the level of >complexity of the organism increases, it will have to do more "information >processing": to find food, to protect against enemies, etc. My point: >intelligence etc. developed out of a need to determine how to behave in >order to survive. So the behaviourist approach is justified: "when the >system seems to act intelligently, it *is* intelligent". You equate intelligence with a high degree of information processing (by co-location of sentences, there is no explicit or clear argument in this paragraph). A cheque clearing system does a high degree of information processing. It must be intelligent then - and AI was achieved 20 years ago? You are making a historical point. Please make it like a competent historian. Otherwise leave evolutionary arguments alone, as you are just making things up. >Before we can write a program for this, we must >understand the algorithm humans use. This proves to be very difficult. >Research is hindered by people claiming that understanding requires very >mysterious causal powers which computers, due to their design, can never have. 'Mysterious' is true only in the sense that we do not yet understand them. 'Eternally mysterious' would not be true. What is true is that causation in human/animal behaviour, and causation in physics, are very different types of cause (explanatory dualism). This does not hold up research at all, it just directs research into different directions. Logical necessity is a further type of pseudo-causation. Its relation to human agency is highly tenuous, and it is wrong to bet too much on it in any research into psychology. Computers cannot uncover mysteries. Automation research may do, in that the task or problem must be properly studied, and it is this study, which advances knowledge rather than the introverted computer simulation. Attempts at computer simulation do, however, expose gaps in knowledge, but this does not make the mystery go away - it only deepens it. The problem is that, if studies are driven by the imperative to automate, this will force the research into an epistemic and methodological straightjacket. This is a narrow approach to the study of human behaviour, and is bound to produce nonsense unless it balances itself with other work. Hence AI texts are far less 'liberal' than psychology ones - the latter consider opposing theories and paradigms. >Gilbert Cockton even claims that because human minds are not >artifacts, while computer systems always will be, there will always be >performance differences. Apart from the fact that this statement is >nonsense, it is not of any importance to AI-research. It is highly relevant. I take it that you think it is nonsense because I offer no support (reasonable) and you don't want to believe it (typical). An artefact is designed for a given purpose. As far as the purpose is concerned, it must be fully understood. The human 'mind' (whatever that is - brain? consciousness? culture? civilisation? knowledge?) was not 'designed' for a given purpose as far as I can see (i.e. I am not a convinced creationist, although nor have I enough evidence to doubt some form of creation). As 'mind' was not designed, and not by us more importantly, it is not fully understood for any of its activities ('brains' are of course, e.g. sleep regulation). Hence we cannot yet build an equivalent artefact until we understand it. Building in itself does not produce understanding. I can expose ignorance, but this is not cured by further building, but by further study. Strong AI does not do this study. My argument is initially from scepticism. I extend the argument to all forms of (pseudo-)intellectual activity which cannot improve our understanding. Strong AI, as modelling without study, i.e. without directed attempts to fill gaps in knowledge by proper, liberal, study, is one such dead-end. Computer modelling based on proper liberal study, is more profitable, but only as a generator of new hypotheses. It does not establish the truth of anything. Finally, establishing the truth of anything concerning human agency, is far far harder than establishing the truth about the physical world, and this is hard enough and getting harder since Quantum interpretations. We have insitutionalised research. There are areas to be studied, and a permanent role in our societies for people who are drawn to advancing knowledge. Unfortunately, too many (of the weaker?) researchers today see any argument on methodological grounds as an attack on research, an attack on their freedom, a threat to the advance of scientific knowledge, a threat to their next funding. The purpose of research is to advance knowledge. Advancing knowledge requires an understanding of what can, and cannot, count as knowledge. In our bloated academia, respect for such standards is diminishing. Research is not hindered by ideas, but by people acting on them. If strong AI cannot win the arguments in research politics, then tough, well - ironic really, for without research politics, it would not have grown as it did in the first place. Those that live by the flam, die by the flam. -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert
smoliar@vaxa.isi.edu (Stephen Smoliar) (03/31/89)
In article <813@htsa.uucp> fransvo@htsa.uucp (Frans van Otten) writes: > >My personal opinion on this is as follows. In the evolutionary process, >with "survival of the fittest", you have to behave in such a way that you >will survive long enough to raise a new generation. As the level of >complexity of the organism increases, it will have to do more "information >processing": to find food, to protect against enemies, etc. My point: >intelligence etc. developed out of a need to determine how to behave in >order to survive. So the behaviourist approach is justified: "when the >system seems to act intelligently, it *is* intelligent". > >Then we invented the computer. We start wondering: can we make this >machine intelligent ? Before we can write a program for this, we must >understand the algorithm humans use. This proves to be very difficult. >Research is hindered by people claiming that understanding requires very >mysterious causal powers which computers, due to their design, can never >have. Gilbert Cockton even claims that because human minds are not >artifacts, while computer systems always will be, there will always be >performance differences. Apart from the fact that this statement is >nonsense, it is not of any importance to AI-research. I find myself basically sympathetic to this approach. However, because recently our Public Television Network has begun a series of programs about current problems in American education, I have been toying with a darker side of this evolutionary model. Let us accept Frans' premise that intelligent behavior emerges because it is necessary for survival (i.e., if you lack physical virtues like strength or speed, you need brains). Then the computer comes along, sort of like the cherry on top of this monstrous technological sundae. At each step in the history of technology, machines have made intelligent behavior less and less necessary for survival Is there a danger that, as machines increase their potential for "intelligent behavior," that they will "meet" the corresponding human potential which is in a decline? Hopefully, this will not be the case. Hopefully, we, as humans, will have to become MORE intelligent in order to interact with the very intelligent machines we build. I just wonder whether or not the technological entrepreneurs who wish to fashion the world in their image will see it that way.
njahren@umn-d-ub.D.UMN.EDU (the hairy guy) (04/01/89)
In article <813@htsa.uucp> fransvo@htsa.uucp (Frans van Otten) writes: >This discussion has degraded into a fight between two groups with different >viewpoints: > > 1. Humans have some mysterious powers that are responsible for their > having a mind. Animals might also have these powers, maybe even > martians. This property might be inherent to the building material; > carbon-hydrogen has it, Si doesn't. > > 2. Understanding etc. are properties which arise from a certain way to > process information. The information theory is what matters, not > the way it is implemented. If we humans can do it using our hardware > (neurons etc), then computers are able to do this using theirs. > I think there is at least one alternative here. In the characterization of (2), I think there is a certain ambiguity in the use of the term "information." The stains on my coffee cup carrie the information that it contained coffee yesterday, but is this information _for_the_coffee_ _cup?_ Certainly not. My mind carries the informatyion that I drank coffee yesterday, and this _is_ information _for_ me. So there is a fundamental difference between two things that we would call information. Now we can ask the question, what is the minimum amount of information that an "information processing" sequence must contain in order to be an instance of mentality. Now one thing that seems reasonable is that the sequence must be able to carry the information that it is _about_ something (other than itself). For instance, if I am thinking about alligators, I must know that I am thinking about alligators, or else I wouldn't be thinking about alligators (this is not a big revelation to most people). Now I might be mistaken, like I might be thinking about an alligator but be attaching the name "crockadile" to it, or I might think that alligators are small furry things that rub up against you, while we can imagine a possible world where alligators are long green things that would just as soon chow on you as look at you. Also, I might not know very much about alligators. For instance, suppose I am a little kid and my mother says to me: "Go tell your father that the alligators are coming up from the swamps and we should leave some milk and cookies for them." Now all I would know about alligators is that they are something my folks are talking about. But in all these cases, I would submit that if I think about alligators, I know that my thoughts are directed towards alligators, it's just that the other mental images of alligators I could appeal to are either incorrect or very sketchy. Now, if we accept this as a pretty necessary feature of mentality, we can ask, in the abstract, whether syntactic digital computation is a sufficiently rich process to carry the information that it is _about_ something. If we find reason to believe thatr it is not, then we would also have reason to believe (1) human mentality is not syntactic digital computation, and (2) syntactic digital computation cannopt give rise to a system of information processing as rich as human mentality, no matter what medium it is implemented in. I see the Chinese Room argument as an argument that Syntactic Digital computation is in fact _not_ sufficiently rich to mee the standard for mentality I oulined above. I personally find it convincing, and would be willing to discuss either this interpretation of CR or the standards above, but I believe I have show that there is at least one reasonable alternative to the two positions Mr van Otten describes above. >order to survive. So the behaviourist approach is justified: "when the >system seems to act intelligently, it *is* intelligent". And all this time I thought there was something called "consciousness." Imagine! But seriously, isn't the question of consciousness and intensionality the question that makes philosophy of mind interesting in the first place. And isn't your behavioristic brushing of them aside tatamount to denying them as important aspects mentality? And if you do choose to deny this, don't you come up with the problem that we _are_ conscious and intensional, and that that's why we're doing all this in the first place? Neal. 530 N 24th Ave E "I cannot disclaim, for my opinions Duluth, MN 55812 *are* those of the Instutute for njahren@ub.d.umn.edu Advanced Partying" "Silence in El Salvador also. Just how old are "the bad old days"? Down goes Vice President Quayle on February 3 to urge good conduct on the Salvadorian Army. On the eve of Quayle's speech, says the human rights office of the Catholic Archdiocese, five uniformed troops broke into the homes of university students Mario Flores and Jose Gerardo Gomez and took them off. They are found the next day, just about the time Quayle and the U.S. press are rubbing noses at the embassy, dead in a ditch, both shot at close range. Gomez's fingernails have signs of "foreign objects" being driven under them. Flores's facial bones and long vertabrae are fractured, legs lacerated, penis and scrotum bruised "as a result of severe pressure or compression." No investigation follows." --Alexander Cockburn _The_Nation_, 3 April, 1989
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (04/02/89)
From article <2705@crete.cs.glasgow.ac.uk>, by gilbert@cs.glasgow.ac.uk (Gilbert Cockton):
" ... As 'mind' was not designed, and not by us more
" importantly, it is not fully understood for any of its activities
" ('brains' are of course, e.g. sleep regulation). Hence we cannot yet
" build an equivalent artefact until we understand it. ...
It doesn't follow. Think of a diamond, for instance.
Greg, lee@uhccux.uhcc.hawaii.edutrevor@mit-amt (Trevor Darrell) (04/02/89)
In article <2705@crete> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: >... >My argument is initially from scepticism. I extend the argument to all >forms of (pseudo-)intellectual activity which cannot improve our >understanding. >... >The purpose of research is to advance knowledge. Advancing knowledge >requires an understanding of what can, and cannot, count as knowledge. >In our bloated academia, respect for such standards is diminishing. Excuse me, but exactly how does one determine when an activity can or cannot improve understanding? And have you published your test of what can, and cannot, count as knowledge? ``References?'' Would you have had all intellectual explorations throughout the ages constrained by these tests? All the artistic explorations? Would you perscribe them as an absolute guide to your child's education? Are you perhaps a bit lacking in the rigor of your debate? (Maybe diatribe is a better term?...) Trevor Darrell MIT Media Lab, Vision Science. trevor@whitechapel.media.mit.edu
ssingh@watdcsu.waterloo.edu ( SINGH S - INDEPENDENT STUDIES ) (04/02/89)
Regarding the division of thought: I think BOTH are right in a way, but both also have their flaws. Carbon and Hydrogen does have something going for it that Silicon does not, that property of being INCREDIBLY plastic. If you could freeze someone at time t, and map out the neural networks, then froze them again at time t +5, you would find changes in the nets. Current computer technology does not allow circuitry to change itself. Unless there is a major revolution in the design of hardware, I honestly think the first truly intelligent computer will be made with organic materials. Who knows? It may even be grown using recombinant DNA or something like that. There is no way we can match the plasticity of the brain with current technology. The second idea is very pure, and very open-ended. While I admire this purity and free-form style, I think it has become undisciplined. Such all encompassing models should be able to express every macroscopic observation in terms of the theoretical models. No one seems to be able to agree on even the most basic of definitions. Plenty of vigour, but no rigour .
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (04/02/89)
It's hard to see where CR understanding might come from (if it exists). It's hard to see where understanding might come from (if it exists). It's hard to see where consciousness might come from (if it exists). It's hard to see where meaning might come from (if it exists). I know why it's hard. These things don't exist. There are theories to the contrary that are embedded in the way we ordinarily talk about people and their behavior. Meaning is the most obvious case. When two sentences are paraphrases, we say they 'mean the same thing'. Then there must be a thing that they both mean, right? Thinking of that as a theory, it might be right, or it might be wrong. We ought to devise alternative theories and look for evidence. We have. I think it's wrong, myself, but opinions differ. If it does turn out to be wrong, then the effort to program meaning into a machine can never be successful -- not because meaning is essentially human, or essentially organic, or essentially analogue, or essentially denotational, or any of the other straws that have been grasped at in this discussion. But because there's simply no such thing in the world to be found in us or to be put into a machine. Greg, lee@uhccux.uhcc.hawaii.edu
murthy@tut.cis.ohio-state.edu (Murthy Gandikota) (04/04/89)
In article <5755@watdcsu.waterloo.edu> ssingh@watdcsu.waterloo.edu ( SINGH S - INDEPENDENT STUDIES ) writes: >+5, you would find changes in the nets. Current computer technology does >not allow circuitry to change itself. Unless there is a major revolution >in the design of hardware, I honestly think the first truly intelligent computer >will be made with organic materials. Who knows? It may even be grown using >recombinant DNA or something like that. There is no way we can match the >plasticity of the brain with current technology. This provokes me to post a thought experiment I've made on self-organizing neural nets. The point is, for a neural net to be as efficient storage/processing device as brain, it should be able to change its connections towards some optimality. Suppose there are two independant concepts A and B represented as two neurons/nodes. So long a relationship is not discovered between them there is no connection between them. Say after some time, a relationship is found between A and B, then a connection can be created between them. However this won't be optimal if A and B have a degree/extent relationship. In which case, A and B have to be merged into some C, with the degrees/extents captured in (the hidden rules of) C. A ready and simple example I can put down is, A=bright red, B=dull red, C=shades of red. Has anyone thought of this before? --murthy -- "What can the fiery sun do to a passing rain cloud, except to decorate it with a silver lining?" Surface mail: 65 E.18th Ave # A, Columbus, OH-43201; Tel: (614)297-7951
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (04/04/89)
In article <3633@uhccux.uhcc.hawaii.edu> lee@uhccux.uhcc.hawaii.edu (Greg Lee) writes: >From article <2705@crete.cs.glasgow.ac.uk>, by gilbert@cs.glasgow.ac.uk (Gilbert Cockton): >" ... As 'mind' was not designed, and not by us more >" importantly, it is not fully understood for any of its activities >" ('brains' are of course, e.g. sleep regulation). Hence we cannot yet >" build an equivalent artefact until we understand it. ... > >It doesn't follow. Think of a diamond, for instance. > Category mistake. Diamonds are a) concrete b) 'assayable' - i.e. you can test chemically that X is indeed a diamond c) synthesisable by following well-understood chemical theories Minds are a) abstract b) not 'assayable' - what the word covers is vague. c) not provably sythesisable becuase of (b) no test for mindhood, and also no theory of how minds get made and function I am still thinking of a diamond however. I cannot think of a mind. -- 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) (04/04/89)
In article <3684@mit-amt> trevor@media-lab.media.mit.edu (Trevor Darrell) writes: > >Excuse me, but exactly how does one determine when an activity can or >cannot improve understanding? And have you published your test >of what can, and cannot, count as knowledge? ``References?'' > I'm utterly derivative :-) Go mug up on epistemology and philosophy of science, then characterise the activities of the various branches of AI (they are not birds of a feather, until you reach the strong AI stuff, like real-world semantics),and then decide to what extent any of them have a coherent view of what it is to promote convincement. >Would you have had all intellectual explorations throughout the >ages constrained by these tests? All the artistic explorations? Would you >perscribe them as an absolute guide to your child's education? Of course not, but we aren't talking about individual exploration, or private enlightenment. We are talking about institutionalised creation of knowledge in the post-war academic culture. This eats resources and shapes people's images of the future. Democracy demands quality control. I know of no incident in the history of science where continued romantic mucking about got anywhere. As A.N. Whitehead argued, all learning must begin with a stage of Romance, otherwise there will be no motivation, no drive to learn, no fascination. But it must be followed by a stage of analysis, a specialisation based on proper discipline, for "where one destroys specialism, one destroys life." With specialism, the mind is "a disciplined regiment" not a "rabble". For Whitehead, a final stage of Generalisation which applies specialised analysis into common sense, real-world context is essential. (A.N. Whitehead, (as in Russel & _) The Aims of Education, 1929) As for children's education, I designed and implemented curricula based on what Whitehead called his 'Rhythm of Education'. AI rarely gets beyond the first beat of the rhythm. Sometimes it dodges it and goes straight into logic etc, but never gets to generalisation, since it was never grounded in anything sensible in the first place. I note that Trevor works in AI vision, which has not got to where it is by romantic sci-fi, but by proper analysis of psychophysical and physiological knowledge. Once reasoning is needed, all this good work goes becomes diluted with the shifting sands of basic AI. -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert
fransvo@htsa.uucp (Frans van Otten) (04/04/89)
Jerry Jackson writes: >This is a seriously flawed statement of the position. It is not that >carbon "has something" that silicon doesn't -- that would be *stupid*. >What is claimed is that possibly it is not merely functional structure >that determines the mind. The "silicon-based" computers we have are >brain-like only in functional organization (if that :-). Perhaps >consciousness is a *chemical* phenomenon and not a result of a particular >functional structure. I don't understand. First you say that it is stupid to believe that carbon has something that silicon doesn't (sounds sensible to me). Then you claim that having a mind might be some chemical phenomenon rather than an abstract information processing phenomenon. So carbon "has it"; "it" would be certain chemical properties. What is the difference ? [...] >I think most people involved in this argument assume that humans evolved >to their present state. This, however, is beside the point ... Searle >and others are simply arguing that something is left out when one does >so ... Humans have a strange attribute known as subjectivity that doesn't >immediately appear to be reducible to structure or functional organization. >It may even be totally unnecessary for intelligent behavior. If so, though, >it is hard to imagine why such a thing would evolve. I don't think that it is beside the point. Evolving to the present state, including the present brain-organization and mind/consciousness/ etc. is a process which is determined by external behaviour. From the evolution point of view, there is no reason why any subjective experience would exist if it would not have a relationship with "the outside world". But such a relationship does exist: when you experience the subjective feeling that you understand, you behave different then when you don't have that feeling. So I claim that *every* subjective experience is an internal state which (partly) determines the behaviour. Gilbert Cockton writes: >>As the level of complexity of the organism increases, it will have >>to do more "information processing" ... My point: intelligence etc. >>developed out of a need to determine how to behave in order to survive. > >You equate intelligence with a high degree of information processing. >A cheque clearing system does a high degree of information processing. >It must be intelligent then - and AI was achieved 20 years ago? Please see the difference between "many simple tasks, all the same" and "many different and difficult tasks". But yes: AI was invented (at least) 20 years ago. The cheque clearing system you write about does understand how to process a check. [...] >>Gilbert Cockton even claims that because human minds are not artifacts, >>while computer systems always will be, there will always be performance >>differences. Apart from the fact that this statement is nonsense, it >>is not of any importance to AI-research. > >An artefact is designed for a given purpose. As far as the purpose is >concerned, it must be fully understood... mind is not fully understood... >hence we cannot yet build an equivalent artefact until we understand it. So when we understand how the human mind works, we can build a machine which has properties like "consciousness", "understanding" etc. Do you claim that this would not be an artifact (maybe because we didn't design it ourselves, but rather copied it) ? Or would we have built an artifact with a mind ? Then there would be no performance differences... That's all I wanted to say. -- Frans van Otten Algemene Hogeschool Amsterdam Technische en Maritieme Faculteit fransvo@htsa.uucp
fransvo@htsa.uucp (Frans van Otten) (04/05/89)
Greg Lee writes: >It's hard to see where CR understanding might come from (if it exists). > >These things don't exist ... the effort to program meaning into a >machine can never be successful ... because there's simply no such >thing in the world to be found in us or to be put into a machine. Without getting too philosophical, let me explain that this is partly true and partly false. Humans are conscious. This is true simply because we state it. But what do we mean by "conscious" ? It is some subjective phenomenon. Subjective phenomena are in my opinion nothing more than certain internal states (a flag that is set, a variable, ...). So when I say "I am conscious" I only say that that specific flag is set. So consciousness does exist (otherwise the word would never have been invented), but it is not a phenomenon that can be observed or detected in the functional structure or in the chemical structure or whereever. The same holds for "understanding". I define "understanding" as "represented in internal symbols". So it is valid to say that my calculator understands addition; it is hard-wired into it. Understanding in itself is useless. But it becomes neccesary when you want to do something with this understood concept (or whatever it is). So when I want to use my calculator, it must be able to perform the rules for addition (which it understands). So I have to feed it with batteries and numbers. At the moment that I feed it with numbers, it understands the instantion of addition I want it to perform now, e.g. 5 + 3 = 8. So "understanding" exists at many levels: I can understand a general concept like addition, but I can also understand an instantion of addition: I understand that 5 and 3 are 8. The more general "rules" that a system "understands" determine the general behaviour of a system. This set of rules is more commonly referred to as the "intelligence" of the system. The more rules the system understands, the more intelligent it is. Don't immediately say "nonsense"; think about it. Don't you ever note the use of this word in general ? In signatures, some people write: Dumb mailers: {backbones}!foo!bar!etc!my_name Intelligent mailers: my_name@etc.foo_domain When we are not discussing AI, "understanding" is a normal word. Under normal circumstances, a watch understands to include februari 29 every fourth year. But when we start discussing AI (or AI-related philosophy) words like "understanding" and "intelligence" get a mysterious load. The Chinese Room Argument is nonsense. Let me explain it once more: 1. It passes the TT. This means: a human being can't tell the difference between a Chinaman and The Room. The behaviour of the room is such that the humans in the jury set their flag: "it understands". 2. John Searle, inside the room, doesn't experience understanding of Chinese. This means: the internal state of the system "Searle" are such that he concludes (unconsciously) not to set his flag "I understand". 3. John Searle then concludes: "As I don't understand, the entire Room doesn't understand" (etc). He should have concluded: "I don't seem to understand, but this doesn't say anything about the Rooms understanding capability". 4. I say: Sure, the room does understand Chinese. Only to the extend that the set of books provide for, of course. And of course, the Room has not the [human] sense of "understanding", as there is no flag for such an internal state. 5. I add to that: In humans, these "flags" are probably located in the right hemisphere. And the symbol grouding problem is probably solved in humans by connecting "understood" concepts to "understood" (physical) sensations. (This latter statement is supported by certain psychological models.) Disclaimer: I am not sure about these statements, they merely seem very probable to me. -- Frans van Otten Algemene Hogeschool Amsterdam Technische en Maritieme Faculteit fransvo@htsa.uucp
cam@edai.ed.ac.uk (Chris Malcolm cam@uk.ac.ed.edai 031 667 1011 x2550) (04/06/89)
In article <2721@crete.cs.glasgow.ac.uk> gilbert@cs.glasgow.ac.uk (Gilbert Cockton) writes: >In article <3633@uhccux.uhcc.hawaii.edu> lee@uhccux.uhcc.hawaii.edu (Greg Lee) writes: >>From article <2705@crete.cs.glasgow.ac.uk>, by gilbert@cs.glasgow.ac.uk (Gilbert Cockton): >>" ... As 'mind' was not designed, and not by us more >>" importantly, it is not fully understood for any of its activities >>" ('brains' are of course, e.g. sleep regulation). Hence we cannot yet >>" build an equivalent artefact until we understand it. ... >> >>It doesn't follow. Think of a diamond, for instance. >> >Category mistake. > >Diamonds are > a) concrete > b) 'assayable' - i.e. you can test chemically that X is indeed a > diamond > c) synthesisable by following well-understood chemical theories > >Minds are > a) abstract > b) not 'assayable' - what the word covers is vague. > c) not provably sythesisable becuase of (b) no test for mindhood, > and also no theory of how minds get made and function There are many different kinds of understanding. People are extremely good at fiddling about and getting things to work, using the minimum understanding necessary for the job. Consequently sailing ships achieved considerable sophistication before the theory of aerodynamics was discovered; and steam engines were made to work not only before the theory of heat engines and thermodynamics existed, but in the face of some wierd and quite wrong ideas about the principles involved. And don't forget that evolution has re-invented the optical eye a number of times, despite never having been to school, let alone having a scientific understanding of optics. Richard Gregory in "Mind in Science" argues that not only do people sometimes make working devices in advance of a proper theoretical understanding of the principles involved, but that this is actually the way science usually progresses: somebody makes something work, and then speculates "how interesting - I wonder WHY it works?" So I expect that AI will produce working examples of mental behaviour BEFORE anyone understands how they work in the analytic sense (as opposed to the follow-this-construction-recipe sense), and that it will be examination and experimentation with these working models which will then lead to a scientific understanding of mind. As for "mind" not being assayable, it's a pity nobody has invented a mind-meter, but we are all equipped with sufficient understanding to be able to say "that looks pretty like a mind to me". Even if closer examination or analysis proves such a judgement wrong, subjecting these judgements to analysis such as the Chinese Room argument, and testing them on the products of AI labs, is a good way of refining them. Current ideas about what constitutes mental behaviour are a good deal more sophisticated than those of several decades ago, partly due to the experience of exercising our concepts of mentality on such devices as the von Neumann computer. I don't see any reason why AI, psychology, and philosophy, shouldn't continue to muddle along in the same sort of way, gradually refining our understanding of mind until the point where it becomes scientific. A (new) category mistake? I assert that I will have a scientific understanding of mind when I can tell you exactly how to make one of a given performance, and be proven right by the constructed device, although such a device had never before been built. Unfortunately I don't expect any of us to live that long, but that's just a technical detail. This idea that you have to understand something properly before being able to make it is a delusion of armchair scientists who have swallowed the rational reconstruction of science usually taught in schools, and corresponds to the notion sometimes held by schoolteachers of English that no author could possibly write a novel or poem of worth without being formally educated in the 57 varieties of figures of speech. It also corresponds to the notion that one can translate one language into another by purely syntactic processing, a notion that AI disabused itself of some time ago after contemplating its early experimental failure to do just that. The human mind is fortunately far too subtle and robust to permit a little thing like not understanding what it's doing to get in the way of doing it. Otherwise we wouldn't even be able to think, let alone create artificial intelligence. -- Chris Malcolm cam@uk.ac.ed.edai 031 667 1011 x2550 Department of Artificial Intelligence, Edinburgh University 5 Forrest Hill, Edinburgh, EH1 2QL, UK
markh@csd4.milw.wisc.edu (Mark William Hopkins) (04/06/89)
In article <820@htsa.uucp> fransvo@htsa.UUCP (Frans van Otten) writes: >Greg Lee writes: > >>These things don't exist ... the effort to program meaning into a >>machine can never be successful ... because there's simply no such >>thing in the world to be found in us or to be put into a machine. > >The same holds for "understanding". I define "understanding" as >"represented in internal symbols". ... and let me add here. Those "symbols" are the symbolic operations that control the body's muscular/skeletal functions. For example, we "understand" the verb to "move" by relating it to the control we exercise over our muscles; the verb to "eat" by our ability to eat and digest food and so on. These are biological universals of the human race that, in effect, create a universal semantic formalism for all human languages; which, in turn, gives us all a common basis for learning our first (and second and ...) language. Understanding a *human* language is intimately related to experiencing our biological condition. BTW, just what is CR anyway?
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (04/06/89)
From article <2721@crete.cs.glasgow.ac.uk>, by gilbert@cs.glasgow.ac.uk (Gilbert Cockton):
" In article <3633@uhccux.uhcc.hawaii.edu> lee@uhccux.uhcc.hawaii.edu (Greg Lee) writes:
" >From article <2705@crete.cs.glasgow.ac.uk>, by gilbert@cs.glasgow.ac.uk (Gilbert Cockton):
" >" ... As 'mind' was not designed, and not by us more
" >" importantly, it is not fully understood for any of its activities
" >" ('brains' are of course, e.g. sleep regulation). Hence we cannot yet
" >" build an equivalent artefact until we understand it. ...
" >
" >It doesn't follow. Think of a diamond, for instance.
" >
" Category mistake.
Whose category mistake? Yours? Certainly not mine. If you had
argued:
As 'mind' was not designed, and not by us more
importantly, and as it is abstract and not 'assayable'
and not provably synthesizable, it is not fully
understood ... Hence we cannot build an equivalent
artifact ...
then I would not have made the particular objection that I made (though
I might have pointed out the curcularity). But that's not what you
said.
" ...
" Minds are
" a) abstract
" b) not 'assayable' - what the word covers is vague.
" c) not provably sythesisable becuase of (b) no test
" for mindhood, and also no
" theory of how minds get made and functionlee@uhccux.uhcc.hawaii.edu (Greg Lee) (04/06/89)
From article <820@htsa.uucp>, by fransvo@htsa.uucp (Frans van Otten): " >no such thing in the world to be found in us or to be put into a machine. " " Without getting too philosophical, let me explain that this is partly " true and partly false. Humans are conscious. This is true simply " because we state it. But what do we mean by "conscious" ? I think there is a crucial confusion in what you say here. The fact that humans say they are conscious is a fact of human behavior, worthy of study. That humans are consious can alternatively be taken as a theory some of us have about human behavior, though perhaps not a very well defined one. Taken as a fact, it's undeniable. Taken as a theory, if it can be made specific enough to have empirical consequences, it can be wrong. I think it is wrong. But more importantly, I think we can't even make sense of these issues, if from the true fact that humans say they are conscious, we draw the conclusion that the (or some) corresponding theory must be correct. That's not sensible. The fact is different from the theory; the theory does not follow from the fact. " ... The Chinese Room Argument is nonsense. ... I'm with you on that point. Greg, lee@uhccux.uhcc.hawaii.edu
gilbert@cs.glasgow.ac.uk (Gilbert Cockton) (04/07/89)
In article <322@edai.ed.ac.uk> cam@edai (Chris Malcolm) writes: >There are many different kinds of understanding. People are >extremely good at fiddling about and getting things to work, using >the minimum understanding necessary for the job. Consequently Oh yawn, yawn, yawn. OK, knowing how, knowing that, knowing why, all different - freshman epistemology this, but don't let me ryle (sic) you :-) >sailing ships achieved considerable sophistication before the theory >of aerodynamics was discovered; and steam engines were made to work >not only before the theory of heat engines and thermodynamics >existed, but in the face of some wierd and quite wrong ideas about >the principles involved. They were making ships, not minds though. If they'd tried to make fish, things would have been different. Ships and steam engines are artefacts with functions, they are not attempts to copy nature. The rest of your argument here falls down because of this simple confusion. Just to make life harder though, "mind" isn't obviously a part of nature in the sense that most people use it (the eternal, that which depends on us not being here). The presence of (the illusion of, for Greg Lee et al.) consciousness, understanding, volition, inspiration and imagination put "mind" out of the normal scope of nature. You're really going to "build" one of these, and say, "well I didn't know what I was doing, but hey I got there, and yeah, it's a mind"? Stop kidding yourself. All examples of artefacts which exploit yet to be undestood phenomena have all been made as tools to solve a narrow problem. This does not apply to strong AI, and your attempt to exploit the analogy does not suggest to me that you have, or indeed want to, think long and hard about the argument. You appear, to me, to be clutching at straws. >mind-meter, but we are all equipped with sufficient understanding to >be able to say "that looks pretty like a mind to me". Sorry, but this is balls - see under animism. There are cultures which give minds to things which ours does not. I suppose they are wrong and we are right? Why not trot down to Portobello at the weekend and ask folk on the beach the following: Which of these have "minds" a) all humans b) Muriel Gray c) Neil Kinnock or Ronald Reagan d) the beach donkeys e) wasps f) slugs g) stroke patients with no mobility or speech h) rivers i) trees j1) god (for atheists) j2) God (for believers) j2 supplementary) the Holy Spirit k) South Bridge l) an intelligent front end to a computer system So what are the right answers? Will we all make the same judgements? If not, then is the Turing Test a sensible basis for a serious research topic? Will the word of one person, enough for Turing, really be enough for research funders? What if ten people are split 8-2 in favour of the computer having a mind? What if the 8 are all children? Come one, let's have some proper criteria for success, not what your mum says. If the Turing Test was THE test, then bring up a wino from Greyfriars today and give him some cans of Heavy to say your robots have minds. Surely that would count as passing the Turing Test? :-) >Current ideas about what constitutes mental >behaviour are a good deal more sophisticated than those of several >decades ago, partly due to the experience of exercising our concepts >of mentality on such devices as the von Neumann computer. Evidence please. I studied cognitive psychology as part of my first degree ten years ago. I presume the progress has happened since then, because things were very poor then. In the late 1970s, there was considerable contempt for the monkeys and cannibals and other artificial problem solving work, which appeared to be the state of the art at the time. What's changed? >This idea that you have to understand something properly before >being able to make it is a delusion of armchair scientists who have You're right that we didn't use to do things this way. However, the idea that you can let loose a artefact which is not understood these days borders on the immoral. Legally, with changes in product liability, you *MUST* undertand what you have made. Hence the social irrelevance of mcuh expert systems work. If they can't be guaranteed, they can't be used. >The human mind is fortunately far too subtle and robust to permit a >little thing like not understanding what it's doing to get in the >way of doing it. Otherwise we wouldn't even be able to think, let >alone create artificial intelligence. What have you been eating? My mind is made, that's how I think with it. Your computer mind isn't made, so you can't think with that. You haven't answered the question of how computer modelling of an ill-defined social construct improves our understanding of nature. -- Gilbert Cockton, Department of Computing Science, The University, Glasgow gilbert@uk.ac.glasgow.cs <europe>!ukc!glasgow!gilbert
fransvo@htsa.uucp (Frans van Otten) (04/07/89)
Greg Lee writes: >Frans van Otten writes: > >>Humans are conscious. This is true simply because we state >>it. But what do we mean by "conscious" ? > >The fact that humans say they are conscious [... can been taken >as] a fact of human behavior [... or as] a theory. Taken as a >fact, it's undeniable. Taken as a theory [...] it can be wrong. >I think it is wrong. But more importantly, I think we can't even >make sense of these issues, if from the true fact that humans say >they are conscious, we draw the conclusion that the (or some) >corresponding theory must be correct. That's not sensible. The >fact is different from the theory; the theory does not follow from >the fact. I can't follow you. What do I mean when I say "I am hungry" ? 1. I am in need of food. 2. I have a (subjective) feeling that I call "hungry". This feeling has been caused by an empty stomach, or by something else. Taken as (1), it is deniable. I can have a hungry feeling without actually needing food. Taken as (2), it is undeniable: I *am* hungry. Maybe this hungry feeling is caused by something else then an actual need for food, but I don't say anything about that. Now when I say "I am conscious", I have a (subjective) feeling which I call "conscious". With this statement, I don't say anything about what might have caused this feeling. In my article, I stated: >>Humans are conscious (1). This is true simply because we state >>it (2). But what do we mean by "conscious" (3) ? Maybe I should have written: (1) Humans say they are conscious. (2) So they have a subjective feeling which they call "conscious". (3) What might this feeling mean, or what might have caused it ? Then I continued my article trying to answer that question. So where do we misunderstand eachother ? -- Frans van Otten Algemene Hogeschool Amsterdam Technische en Maritieme Faculteit fransvo@htsa.uucp
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (04/08/89)
From article <322@edai.ed.ac.uk>, by cam@edai.ed.ac.uk (Chris Malcolm cam@uk.ac.ed.edai 031 667 1011 x2550): " ... " This idea that you have to understand something properly before " being able to make it is a delusion of armchair scientists who have " swallowed the rational reconstruction of science usually taught in " schools, and corresponds ... It also corresponds to the notion " that one can translate one language into another by purely syntactic " processing, a notion that AI disabused itself of some time ago after " contemplating its early experimental failure to do just that. What you say seems to assume that the syntax of natural languages was or is understood. That is not the case. It's very far from being the case. The failure you mention, consequently, does not suggest that translation cannot be achieved by syntactic processing. Greg, lee@uhccux.uhcc.hawaii.edu
lee@uhccux.uhcc.hawaii.edu (Greg Lee) (04/08/89)
From article <826@htsa.uucp>, by fransvo@htsa.uucp (Frans van Otten): " ... " I can't follow you. What do I mean when I say "I am hungry" ? " " 1. I am in need of food. " 2. I have a (subjective) feeling that I call "hungry". This " feeling has been caused by an empty stomach, or by something " else. " " Taken as (1), it is deniable. I can have a hungry feeling without " actually needing food. Taken as (2), it is undeniable: I *am* " hungry. Though I have reservations about (2), I think this is a good analogy. One might say "I am hungry" to mean "Give me some food now" or to announce an intention to raid the refrigerator. If so, it's an explanation of the demand or of the up-coming behavior. As students of human behavior, we might take the explanation seriously, and go about the business of trying to put it in a rigorous way by trying to identify a chemical hunger-need syndrome or a neuronal hunger-feeling syndrome. That's step one. Having done that, as step two, we could investigate its truth. Maybe for hunger-feeling it would have to be true, as you say. I don't think that's so clear. But what about step one? Does it have to succeed? We're dealing with a folk explanation of behavior. Maybe it's just wrong. Similarly consiousness. " Maybe this hungry feeling is caused by something else " then an actual need for food, but I don't say anything about that. " " Now when I say "I am conscious", I have a (subjective) feeling which " I call "conscious". With this statement, I don't say anything about " what might have caused this feeling. That's not what I meant by saying it was a theory. I meant that statements about consiousness are used ordinarily to explain or justify behavior. Folk theories are advanced. " ... So where do we misunderstand eachother ? Since you don't recognize the theoretical nature of humans' statements about consciousness, you are prevented from entertaining the possibility that the theories might be incorrect. Greg, lee@uhccux.uhcc.hawaii.edu