cw2k+@andrew.cmu.edu (Christopher L. Welles) (11/08/90)
I've just finished a paper on Searle's "Minds, Brains, and Programs." Just thought I'd post it to see what other people thought. Anyone agree with me?? Brains, the Magic Vessel of Intentionality In his "Minds, Brains, and Programs", John Searle's whole conclusion is based on a bad assumption. The responses to his paper do bring up some things worth considering, but this occurs only because Searle's statements were misinterpreted. His ideas, themselves, represent something that is obviously, and completely wrong, not worthy of consideration. His whole paper represents a lack of understanding of how formal systems work, what "intentionality" is, and the difference between hardware and software. Beginning with his first example, Searle makes a tragicly false assumption. Searle claims that since the English speaking man, who performs all the formal operations, does not understand Chinese, there is no understanding involved. The obvious problem with this, is the Systems Reply: The system as a whole understands. The man is only a part of the system, therefore he does not understand. This was understandably misinterpreted by Searle. As he understood it, it implied that simply internalizing the system( i.e. memorizing the rules, script, etc. ) would mean that the English speaking man must therefore understand Chinese. Something that would obviously not be true. The point of the Systems Reply, however, has nothing to do with the physical representation of the system. For all it cares, the components of the system could be spread all over the world. It depends Foley upon the logical structure, and how they interact. The system. Stated more clearly, the English speaking man was only performing the part of a machine, executing a program. It is the result which understands, and not the machine itself. In order for the English speaking man to perceive understanding in the system, he would have to see it from the system's viewpoint. Not just performing the actions that bring the system about. An analogy to this would be a neuron in the brain. It merely reacts according to specified rules, a program. Even if it were capable, it would have no understanding of what it's actions represented to the whole system, i.e. the mind. Pulses are just received and sent according to a set of rules. You can apply this to the whole brain for that matter. The brain itself does not understand, only the result of it's execution, the mind. To say otherwise, you would have to say that somehow, if the brain stopped executing it's instructions, there would still be understanding. After all, if the understanding were inherent in there simply being a brain, why bother acting according to fixed rules. Do you decide to keep following those fixed rules so you will continue thinking??? The idea that following formal principles will cause you to understand is absurd. Do you feel that you keep following a set of rules so that you keep understanding? You are, however, a result of formal rules being followed. Your brain, acts according to physical principles, a set of rules. This distinction between mind, is exactly one of the things Searle argues against: "The distinction between the program and its realization in hardware seems to be parallel to the distinction between the level of mental operations and the level of brain operations. . . . But the equation "mind is to brain as program is to hardware" breaks down. . . . the distinction between program and realization has the consequence that the same program could have all sorts of crazy realizations. . . . Weizenbaum, for example, shows in detail how to construct a computer using a role of toilet paper and a pile of small stones. Similarly, the Chinese. . . program can be programmed into a sequence of water pipes, a set of wind machines, or a monoligual English speaker, none of which thereby acquires understanding of Chinese."(Perry, p. 400-1) All this means, the only thing he demonstrates throughout his whole paper, is that performing a series of rules can not result in one understanding. However, it does not mean that no understanding will result, just that the one executing the program won't be the one doing it. There is nothing against this idea anywhere in AI. No one expects the circuits to begin understanding just because it's following a certain program. A program, in the state of being executed, might be said to be understanding however. It seems Searle never got this distinction strait however: "One of the claims made by the supporters of strong AI is that when I understand a story in English, what I am doing is exactly the same--or perhaps more of the same--as what I was doing in manipulating the Chinese symbols."(Perry, p. 393) Searle obviously didn't understand what was meant here. In the case of him performing symbol manipulation, yes that is done. But it is not something he, his understanding conscious self, does. The same term is being used here to actually refer to two different things. Both his mind, and his body; the brain being considered part of the body. There is generally no need to distinguish between them because, at least at our current technology level, the mind and the body happen to stay together. Searle's argument against the Chinese box is simply that there can be no understanding with just formal symbols. The symbols must have some intrinsic meaning assigned to them. He terms this "intentionality." By his own definition, intentionality is "that feature of certain mental states by which they are directed at or about objects and states of affairs in the world." While this, at least seems to hold with the generally accepted idea of what intentionality, Searle's interpretation of it distinguishes it. When Searle says something has intentionality, he means that it has an intrinsic meaning of some sort. Some have tried to take this to mean that the symbol and what it's referring to are causally connected in some way, but Searle's reply clearly rules that out: "no matter what outside causal impacts there are on the formal tokens, these are not by themselves sufficient to give the tokens any intentional content. No matter what caused the tokens, the agent still doesn't understand Chinese"(The Behavioral and Brain Sciences, p. 454) By this, and other comments, intentionality seems to be some method of letting the brain know, without referring to other symbols, what the symbols coming in mean. Sort of like a dictionary entry for each symbol coming in, but instead of defining it in words( themselves symbols ), it simply carries the meaning: "for intentionality there must be intentional content in addition to the formal symbols."(The Behavioral and Brain Sciences, p. 454) Of course, how do you communicate meaning without referring to symbols? This is the very thing that Searle supposes biological material is capable of, and that's why we can't make a program that involves, in and of itself, understanding. It depends upon the material to contain the meaning. Lets look at a description of how we see things according to Searle: "From where I am seated, I can see a tree. Light reflected from the tree in the form of photons strikes my optical apparatus. This sets up a series of sequences of neural firings. Some of these neurons in the visual cortex are in fact remarkably specialized to respond to certain sorts of visual stimuli. When the whole set of sequences occurs, it causes a visual experience, and the visual experience has intentionality. It is a conscious mental event with an intentional content."(The Behavioral and Brain Sciences, p. 452) It follows an expected sequence at first, but then, after the neuron firings, is suddenly translated into something with intentionality. Notice that, it is no longer represented by symbols(i.e. the firing of the neurons) but is instead, by the great power of organic material, now carried around the brain as an image. His whole ridiculous idea of intentionality, and the idea that somehow the meaning must be passed on to that agent carrying out the program hinges on the idea that the understanding is done by the agent. What Searle is trying to do, is set up a Chinese box wherein, the meaning of the Chinese symbols, are translated into English and the Person inside the box does the understanding. That, however, only leads to breaking it down further and further. Is there sum sub-atomic particle that does the understanding for us. --------------------------------------------------------------- Hope you liked it, or maybe hated it. Let's hear some opinions! <<<<< Chris >>>>>
larryc@poe.jpl.nasa.gov (Larry Carroll) (11/09/90)
In article <deleted> cw2k+@andrew.cmu.edu (Christopher L. Welles) writes: >Searle argues against: > >"The distinction between the program and its realization in hardware >seems to be parallel to the distinction between the level of mental >operations and the level of brain operations.... But the equation "mind is >to brain as program is to hardware" breaks down.... the distinction >between program and realization has the consequence that the same >program could have all sorts of crazy realizations.... Weizenbaum, for >example, shows in detail how to construct a computer using a roll of toilet >paper and a pile of small stones. Similarly, the Chinese... program >can be programmed into a sequence of water pipes, a set of wind machines, or >a monolingual English speaker, none of which ... It would be interesting to ask Searle if he thinks that an alien creature with a base in an entirely different biology could be intelligent or have consciousness. Say, fluorine-silicon rather than hydrogen-carbon, using thermal energy reactions that mimic our form of oxidation. Or something even more radical: plasma life-forms living in a stars chromosphere, using fusion reactions rather than chemical reactions. Larry Carroll "Takes-us" (correct pronunciation of Texas) Dancin' Fool
sarima@tdatirv.UUCP (Stanley Friesen) (11/10/90)
In article <kbCADGe00VsL5230Uu@andrew.cmu.edu> cw2k+@andrew.cmu.edu (Christopher L. Welles) writes: >I've just finished a paper on Searle's "Minds, Brains, and Programs." >Just thought I'd post it to see what other people thought. Anyone agree >with me?? ... > An analogy to this would be a neuron in the brain. It merely >reacts according to specified rules, a program. Even if it were capable, it >would have no understanding of what it's actions represented to the whole >system, i.e. the mind. Pulses are just received and sent according to a >set of rules. ... This is very critical. There is *no* part of the brain which understands, only the whole brain, operating normally can be said to understand. Certainly any given part of the brain, when examined in detail, performs a rather rigid, relatively simple data transformation. This has been true for *every* brain region so far studied. The cerebellum is a particularly good example. Any given region of the cerebellum performs an *identical* tranformation, yet due to its external wiring it may perform any of several functions, including balance, muscle tone, motion smoothing, and perhaps others. Other section of the brain are similar in that they are data transformers that do not 'care' where the data came from or where it is going. [Of course this whole discussion applies to individual neurons in spades] ... > By this, and other comments, intentionality seems to be some >method of letting the brain know, without referring to other symbols, >what the symbols coming in mean. Sort of like a dictionary entry for each >symbol coming in, but instead of defining it in words( themselves symbols >), it simply carries the meaning: "for intentionality there must be >intentional content in addition to the formal symbols."(The Behavioral >and Brain Sciences, p. 454) > Of course, how do you communicate meaning without referring >to symbols? This is the very thing that Searle supposes biological material >is capable of, and that's why we can't make a program that involves, in >and of itself, understanding. It depends upon the material to contain the >meaning. Lets look at a description of how we see things according to >Searle: Oh wonderful :-) And just what does he see in brains that provides this non-referential semantics? How do patterns of neuronal firing differ from any other symbolic system? [A pattern of neuronal firings is as much a symbol as anything else, it has no particular relationship to any meaning it might have] As far as I can see Searle's basic problem is a lack of understanding of neurobiology. He seems to have no concept of the current state of neural science, especially with regard to how a brain processes data. >"From where I am seated, I can see a tree. Light reflected from the tree in >the form of photons strikes my optical apparatus. This sets up a series of >sequences of neural firings. Some of these neurons in the visual cortex are >in fact remarkably specialized to respond to certain sorts of visual stimuli. >When the whole set of sequences occurs, it causes a visual experience, and >the visual experience has intentionality. It is a conscious mental event >with an intentional content."(The Behavioral and Brain Sciences, p. 452) >It follows an expected sequence at first, but then, after the neuron >firings, is >suddenly translated into something with intentionality. Notice that, it is >no longer represented by symbols(i.e. the firing of the neurons) but is >instead, by the great power of organic material, now carried around the >brain as an image. Oh foo!! This is starting to get totally silly. I have a hard time believing anyone, even Searle could really believe this! What does he think a 'visual experience' is if it is not a pattern of neuronal firing?? Jeez! [So some of the earlier neuron in the chain have a very specific response mode, most of these are essentially the preprocessor that converts light into an internal mental model (i.e. into symbolic form). It is not until the more complex cells of the tertiary visual areas are reached that we can talk about symbols in any meaningful sense. But even there everything is still a pattern of neuronal firing.] And what in blue blazes is an 'intentional content'? If he just means that the symbols have a behavioral significance, so what? If he means something else, what is its physical basis in brain activity? [And if it *has* no physical basis, he is just reinventing dualism] > His whole ridiculous idea of intentionality, and the idea that >somehow the meaning must be passed on to that agent carrying out the >program hinges on the idea that the understanding is done by the agent. And it thus becomes a problem of infinite regress - where is the 'agent' in the human brain? How does a brain qualify as an 'agent'? >... That, however, only leads to breaking it down >further and further. Is there sum sub-atomic particle that does the >understanding for us. Quite, and even then, who's to say we have reached the end??? As you can see I basically agree with you. And I am coming from a background in the biological sciences. Searle is a biological idiot. My main problem with AI as a field is that I think most research is approaching the problem bass-ackwards. Too much time is being spent on finding 'reliable' reasoning paradigms for artificially contrained, highly unnatural domains. This is *not* how natural intelligence works. -- --------------- uunet!tdatirv!sarima (Stanley Friesen)
fraser@bilby.cs.uwa.oz.au (Fraser Wilson) (11/10/90)
In <10297@jpl-devvax.JPL.NASA.GOV> larryc@poe.jpl.nasa.gov (Larry Carroll) writes: >It would be interesting to ask Searle if he thinks that an alien creature >with a base in an entirely different biology could be intelligent or have >consciousness. Say, fluorine-silicon rather than hydrogen-carbon, using >thermal energy reactions that mimic our form of oxidation. Or something even >more radical: plasma life-forms living in a stars chromosphere, using fusion >reactions rather than chemical reactions. Searle's argument had nothing to do with conciousness residing in alternative biologies. What he was essentially saying was that a _formal system_ (which is what a computer is) can never be conscious. I think the systems reply essentially chucks this one right out the window where it belongs :-). > Larry Carroll > "Takes-us" (correct pronunciation of Texas) > Dancin' Fool Fraser Wilson. (fraser@bilby.cs.uwa.oz)
miodeen@buddha.ncc.umn.edu (Mike Odeen) (11/11/90)
I've always felt that Searle's argument falls apart when you deny that anything like "meaning" or "intentionality" actually exist. Meaning results from an interaction between symbols. Look at a dictionary. It wont give you the meaning of any word, but it will provide you with other words, setting up an association between them. No one element carries any meaning, just an association value with other elements. Something like what happens in the dictionary caould easily be instantiated in some kind of network model of the brain. "Meaning" if it exists anywhere would just be a complex interaction between symbols in the brain. -- Michael J. Odeen miodeen@buddha.ncc.umn.edu
smoliar@vaxa.isi.edu (Stephen Smoliar) (11/12/90)
In article <10297@jpl-devvax.JPL.NASA.GOV> larryc@poe.jpl.nasa.gov (Larry Carroll) writes: > >It would be interesting to ask Searle if he thinks that an alien creature >with a base in an entirely different biology could be intelligent or have >consciousness. I seem to recall seeing him address this point in one of his many publications on this subject. As I recall, he takes the rather dubious stand that ANY chemical basis for life may at least potentially possess his precious intentionality. This means that he is essentially saying that chemical processes may ultimately give rise to intentionality while computational ones cannot. This does not strike me as particularly sound, but then I am not sure that I can claim to really understand just what Searle has in mind when he brings intentionality into the argument. (Perhaps he just wishes to use it as a cross to ward off those vampires who actually write AI programs!) ========================================================================= USPS: Stephen Smoliar 5000 Centinela Avenue #129 Los Angeles, California 90066 Internet: smoliar@vaxa.isi.edu "It's only words . . . unless they're true."--David Mamet
shafto@ils.nwu.edu (Eric Shafto) (11/12/90)
In article <57@tdatirv.UUCP> sarima@tdatirv.UUCP (Stanley Friesen) writes: > This is very critical. There is *no* part of the brain which understands, > only the whole brain, operating normally can be said to understand. I think this still leaves you vulnerable to Searle's argument. The critical fact is this: Not only does no part of the brain understand, the brain doesn't understand. The system of which the brain is a part (for which the brain is the substrate?) understands. Let's call that system the mind. In other words, *I* understand, but my brain does not. Am I my brain? I think not. The system in the room understands, but the human does not. Where's the contradiction? Searle's genius lies in making things seem overwhelmingly counterintuitive. He never actually proves or disproves anything. Regards, Eric Shafto Institute for the Learning Sciences Northwestern University
raphael@fx.com (Glen Raphael) (11/13/90)
cw2k+@andrew.cmu.edu (Christopher L. Welles) writes: >I've just finished a paper on Searle's "Minds, Brains, and Programs." >Just thought I'd post it to see what other people thought. Anyone agree >with me?? I agree with you that the Chinese room is a lot of hooey; wrote a 20-page pager on it myself once... >The responses to his paper do bring >up some things worth considering, but this occurs only because Searle's >statements were misinterpreted. Not at all; there just happen to be a LOT of ways in which to attack the Chinese Room concept, most of which are valid... >His whole paper represents a lack of understanding of how >formal systems work, what "intentionality" is, and the difference between >hardware and software. I don't think you should accuse Searle of misunderstanding intentionality unless you are willing to attack *his* concept of intentionality at length; he wrote a book called "Intentionality" which is a standard text in Speech Theory. >), it simply carries the meaning: "for intentionality there must be >intentional content in addition to the formal symbols."(The Behavioral >and Brain Sciences, p. 454) Again, for a full formal attack from this angle you'll have to look at how Searle defines "intentional content". I found it fairly easy to attack the Chinese room by reductio ad absurdum based on a few thought experiments. For furthur ideas in this vein you might read the answer to Searle in "The Mind's I" by Douglas Hofstadter and Daniel Dennett. The Chinese Room is basically an "argument from intuition" rather than a proof, and most of us on the net probably have a contrary intuition. Searle gets a lot of mileage out of a conviction that basically boils down to "If computers could think, then systems made out of rocks and toilet paper, or even *beer cans* could think, and that's PREPOSTEROUS!" The rest is handwaving. (I'll have to take another look at my paper one of these days...) Incidentally, Searle is a *fantastic* lecturer; he's just out of his depth when it comes to AI. > <<<<< Chris >>>>> Glen Raphael raphael@fx.com
cw2k+@andrew.cmu.edu (Christopher L. Welles) (11/13/90)
In <10297@jpl-devvax.JPL.NASA.GOV> larryc@poe.jpl.nasa.gov (Larry Carroll) writes: >It would be interesting to ask Searle if he thinks that an alien creature >with a base in an entirely different biology could be intelligent or have >consciousness. Say, fluorine-silicon rather than hydrogen-carbon, using >thermal energy reactions that mimic our form of oxidation. Or something even >more radical: plasma life-forms living in a stars chromosphere, using fusion >reactions rather than chemical reactions. In<fraser.658229808@bilby>, fraser@bilby.cs.uwa.oz.au (Fraser Wilson) writes: >What he was essentially saying was that a >_formal system_ (which is what a computer is) can never be conscious. >I think the systems reply essentially chucks this one right out the >window where it belongs :-). Just thought I should point out. Searle did make it clear that "formal systems", computers, could be conscious. He emphasized the fact that humans were such systems. His main thesis went something to the effect of: That there mere instantiation of a program could not be in itself, sufficient for for consciousness. In this, he is saying that consciousness somehow depends upon the "stuff" that the computer is made out of. In fact, he did claim that, Martians, for example, might have consciousness, as well, but it would depend upon the "stuff" they were made out of. Somehow, conscious systems, unlike another formal system, could not be represented in any symbolic structure. The material had to have this special property of intentionality. Something, he theorized, only maybe human brains had. Throughout Searle's Chinese room paper, it is easy to misunderstand what Searle is saying. Just the fact that what he's saying really doesn't make any sense, causes one to read into it something completely different. A major problem with the commentaries is that almost no one really understood all that Searle was saying. Many of the arguments there were designed to deal with the point they thought Searle was trying to make. What he actually said was just too ridiculous. Only after reading his response to several of the commentaries can it really understood what Searle is saying. When I first read his Chinese room paper, itself, I read it three times just to figure out what he was saying about intentionality. Even then it wasn't clear though. Only after reading some commentary on it, as well as Searle response to the commentary did I have any chance of figuring out what he was really saying. I usually assume that the writer is a rational human being. It looks as if that just doesn't apply for Searle. I suppose that last isn't really true, it just that Searle had absolutely no concept of how a computer, or a computer program actually worked. It's obvious that Searle was arguing about a subject in which he had absolutely no background. Anyone, dealing either with computers, or (is it operational psychology?) would have immediately seen such an idea was wrong. <<<<< Chris >>>>>
cam@aipna.ed.ac.uk (Chris Malcolm) (11/14/90)
In article <10297@jpl-devvax.JPL.NASA.GOV> larryc@poe.jpl.nasa.gov (Larry Carroll) writes: >It would be interesting to ask Searle if he thinks that an alien creature >with a base in an entirely different biology could be intelligent or have >consciousness. He answered that in 1980 in the replies to his original BBS target article criticism, when he said: We need to keep reminding ourselves over and over: Cognition is a biological phenomenon - as biological as digestion, photosynthesis, lactation, or the secretion of bile. We might do any of one of these in an artificial medium removed from normal biochemistry, but we couldn't do any one of them by pure syntax. Somewhere else I can't remember he quite explicitly said that a computer couldn't be made to think, but a machine could -- after all what were we but biological machines? So he certainly doesn't deny the possibility of cognition to robots or Martians, just automated syntax shufflers. I don't mean to attack Larry Carroll here, but there are too many posters in this thread who haven't read much (or any) Searle, quite apart from the young twits who say "I haven't read any of ... but it seems to me ... etc." If you want to know what Searle thought don't read this newsgroup! Read Searle! He's not as stupid as most of those who disagree with him :-) -- Chris Malcolm cam@uk.ac.ed.aipna 031 667 1011 x2550 Department of Artificial Intelligence, Edinburgh University 5 Forrest Hill, Edinburgh, EH1 2QL, UK
cam@aipna.ed.ac.uk (Chris Malcolm) (11/14/90)
In article <57@tdatirv.UUCP> sarima@tdatirv.UUCP (Stanley Friesen) writes: >My main problem with AI as a field is that I think most research is >approaching the problem bass-ackwards. Too much time is being spent on >finding 'reliable' reasoning paradigms for artificially contrained, >highly unnatural domains. This is *not* how natural intelligence works. I agree. Most AI research has been doing this. But not all. Moravec suggested in "Locomotion, Vision, and Intelligence", in Robotics Research 1, eds Brady and Paul, MIT Press, 1984 that "... the most fruitful direction for this track [the development of artificial intelligence] is along the oxtrail forged by natural evolution before stumbling on us ... developing a responsive mobile entity ...". And there are some (e.g. Minsky) who were never much taken with the logical approach. Six years on there's quite an entertainingly active research field. Not to mention the confectionists YAWNs (Yet Another Wonderful Network) and sundry other PDP approaches of various granularities. The big money and the textbooks are still pushing Liebniz's dream of finding the calculus of thought, but there's also a lot of diverse research programmes which say, like you, that "is *not* how natural intelligence works". -- Chris Malcolm cam@uk.ac.ed.aipna 031 667 1011 x2550 Department of Artificial Intelligence, Edinburgh University 5 Forrest Hill, Edinburgh, EH1 2QL, UK
blenko-tom@cs.yale.edu (Tom Blenko) (11/15/90)
In article <3488@aipna.ed.ac.uk> cam@aipna.ed.ac.uk (Chris Malcolm) writes: | |If you want to know what Searle thought don't read this newsgroup! Read |Searle! He's not as stupid as most of those who disagree with him :-) Well said. I have for some time regarded Searle's original article as an intelligence test for the AI community (it is even more entertaining to suppose that that was its intent). Not because there's any a priori reason it should serve as such, it just it seems to work pretty well that way. Tom
me@csri.toronto.edu (Daniel R. Simon) (11/15/90)
Followers of the current debate in this newsgroup on Searle's "Chinese room" argument, who are familiar, through my last posting here, with the Laboratory for Artificial Appearance, may be interested in an email message I received from a friend and contact there not long ago. [For those who are unfamiliar with the LAA, it was originally the graphics group within the computer science department of a university that shall remain nameless. It owes its inspiration (and its new name) to a revolutionary paper, published some time ago in an obscure graphics journal, which introduced the concept of "artificial appearance" by proposing the following test: a room is fitted with two high-resolution CRT monitors, one of which is connected to a camera pointed at a real person in another room, and the other of which receives its input from a computer graphics generation program. If a viewer is unable to tell, by scrutinizing the two screens, which one displays the image of a real human being, the paper's author argued, then there is no reason not to assume that the computer has artificially generated human appearance. This so-called "blurring test" has revolutionized thinking about computer graphics, prompting heated philosophical debates about the nature of human appearance, and radically altering prevailing opinions on the best approaches to its artificial simulation.] Here is the message I received: Hi, Dan! I hope you enjoyed your visit with us, and that your next visit will be soon. Several people here at LAA have remarked to me that they found their discussions with you to be most fruitful. I thought I'd bring you up to date on recent developments since you left. At that time, the mood here was (as I expect you perceived) fairly confident among us; we felt (and still feel) that while we have only begun to make a dent in the huge problems involved in creating artificial human appearance, we have at least been making steady progress towards that goal. Imagine our surprise when a well-respected local professor of philosophy here began, just a few days later, distributing advance copies of a soon-to-be-published article he wrote, arguing that artificial human appearance is not merely difficult to achieve, but actually impossible in principle! Our surprise quickly turned to scorn, however, once we were able to study his arguments in detail. His main premise is that human appearance is inherently "continuous", and cannot be simulated by any "discrete" representation, such as a computer would generate. He illustrates his thesis as follows: suppose that he were to possess the entire text of such a representation (in, say hexadecimal code) for the appearance of a 21-year-old Chinese man in traditional wedding garb. Suppose further that a particular video monitor receives its input not from a camera or digital input, but from a device which turns short, sharp tapping sounds into numbers (the way telephone exchanges do when old-style "pulse" telephones are dialed). Then, merely by tapping on the table (or on himself, if necessary), he can, in principle, conjure up on this monitor the image of said 21-year-old Chinese man. "Does that mean", the author asks provocatively, "that I look Chinese?" I expect I need hardly point out to you the obvious flaw in this "Chinese groom" argument. It is clearly not the man himself, but the whole array of person, text, and monitor that realizes the appearance of the Chinese man. The author himself even recognizes the possibility of this refutation, admitting that the presence of the text may make his claim of Chinese appearance suspect; he replies with a revised scenario in which he has memorized the entire text. However, he never follows the objection to its logical conclusion, recognizing that the whole combination of "hardware" and "software" can be said to be forming an artificial human appearance. In retrospect, I find it somewhat puzzling that his paper has garnered so much attention, given the weakness of its arguments. I'm interested to hear your opinion, of course, but I'd be surprised if you lent his propositions any more credence than we have. Anyway, keep in touch, and let me know when you'll be in the area again. D----. (I include here my reply to him:) Well, I certainly agree with you that the philosophy professor's argument seems weak--after all, surely there is some level (molecular? subatomic?) at which the real appearance of a real person is either discrete or negligibly different from a discrete representation of it. On the other hand, I have some nagging questions about your willingness to localize appearance in what seems to me to be a rather arbitrary way. If a person, a text, an electronic apparatus, and a monitor can be said to have a human appearance, then what about the person, the text, the apparatus, the monitor, and the stand on which the monitor rests? Or all of these things, plus me? Do two people look like one person? Does a volleyball team? The earth, including all its inhabitants? I get the feeling that if I were compelled to localize human appearance (at least as defined by the "blurring test"), I would do so in an altogether different way. But I suppose I'd best defer to the experts on that score. Anyway, best of luck with your research, and I hope to visit again some time soon. "There *is* confusion worse than death" Daniel R. Simon -Tennyson (me@theory.toronto.edu)
fostel@eos.ncsu.edu (Gary Fostel) (11/17/90)
I always wondered why people spent so much time argueing about the Turing
Test when it is so poorly defined. (Or perhaps that's why it can go on and
on). Consider a modification for a chess playing version of the Turing
test. (We could use Chinese Chess if that would help :-)
My wife would probably be greatly impressed by almost any computer chess
playing program and would be forced to conclude that BY HER STANDARDS,
the computer was passing any reasonable test she could construe of the
machines ability to play chess *as* a human would. But of course, she
knows very little of chess or algorithms for chess and implementations of
same on computers. I know a bit more about each and I would be able to
detect some properties that might give it away, e.g. very constant
response time, or pathological weaknesses in certain features if its
play, e.g. with non-book openings or cleverly arranged combinations of
various lengths to detect its search depth etc. Yes, I know there are
chess programs that would pass these tests, but there are people more
expert than me who would be able to contruct more useful tests for
these "smarter" programs.
The only well defined "test", whether it be for chess playing or more
generally, in the Turing test, for intelligence, would be that the machine
could fool EVERYBODY who was observing. This is a very strong test,
much stronger than the usual Turing test I believe, and I'm not sure
many REAL people could convince everyone else that they were intelligent
humans. Perhaps none could!
So ... if I try to make the Turing test into a definite, well defined
test, it is not clear that ANYTHING could ever pass it.
There's no doubt that the Turing test is great sport, but when
you come right down to it, it is not very useful. If I follow this one
step, it means that the Chinese Room device is also not very useful since
it exists merely to weaken another device (the Turing Test) which
can be weakened far more without recourse to the Chinese Room -- which
muddies the situation rather than clarifying it.
----GaryFostel---- Department of Computer Science
North Carolina State University
tj@pons.cis.ohio-state.edu (Todd R Johnson) (11/17/90)
In article <3488@aipna.ed.ac.uk> cam@aipna.ed.ac.uk (Chris Malcolm) writes: >>If you want to know what Searle thought don't read this newsgroup! Read >>Searle! He's not as stupid as most of those who disagree with him :-) Agreed. After disagreeing with Searle I went back and re-read (and re-read) his Scientific American article. I still don't agree (or perhaps understand) the point he is trying to make with the Chinese Room argument. However, his main point seems to be that any intelligent artifact with the capabilities of a human must be built in hardware that is at least as power as the human brain. In other words, if we are to produce intelligent artifacts we MUST be willing to accept the fact that we need to create special software AND hardware. This seems quite reasonable. In fact, I don't see how anyone can disagree with it. ---Todd -- Todd R. Johnson tj@cis.ohio-state.edu Laboratory for AI Research The Ohio State University
smoliar@vaxa.isi.edu (Stephen Smoliar) (11/18/90)
In article <3488@aipna.ed.ac.uk> cam@aipna.ed.ac.uk (Chris Malcolm) writes: > >If you want to know what Searle thought don't read this newsgroup! Read >Searle! He's not as stupid as most of those who disagree with him :-) This is an important point, but there is also a major CAVEAT. Listening to Searle speak should not be regarded as an adequate substitute for reading his publications. When I heard him at UCLA, he indulged in a nasty habit of playing to his audience, putting far more emphasis on making his phrases dance than on conveying the content of his message. The best way to deal with Searle is on the printed page (and in the sort of relaxed mood which is likely to open the mind). I also feel that it is about time we apply this new rule to Turing as well. I cannot help but wonder how many participants in this debate (mind you, that includes Searle) have actually READ "Computing Machinery and Intelligence," the source of that now notorious test which continues to inspire such controversy. What is particularly important is that, in his first paragraph, Turing recognizes that "think" is probably too highly-charged a word to serve as a basis for discussion. The whole purpose of what is now known as the Turing test was to recognize that "Can machines think?" was too vague a question and to replace it with one which could be "expressed in relatively unambiguous words." If Turing is now looking down on us from Heaven, he is probably aghast at all the ways in which his simple intellectual exercise has been abused. ========================================================================= USPS: Stephen Smoliar 5000 Centinela Avenue #129 Los Angeles, California 90066 Internet: smoliar@vaxa.isi.edu "It's only words . . . unless they're true."--David Mamet
cam@aipna.ed.ac.uk (Chris Malcolm) (11/19/90)
In article <8bDqHlK00VsLBAOkxp@andrew.cmu.edu> cw2k+@andrew.cmu.edu (Christopher L. Welles) writes: >Just thought I should point out. Searle did make it clear that "formal >systems", computers, could be conscious. He emphasized the fact that >humans were such systems. No. He said that machines could underatand, that we were no more than biological machines, but denied that a computer running any program would be able to do this, on the grounds that in this last case ALL that is going on is syntactical, and you can't get semantics into a syntactic process by whatever elaborations of syntactic manipulation. >His main thesis went something to the effect of: That there mere >instantiation of a program could not be in itself, sufficient for for >consciousness. In this, he is saying that consciousness somehow depends >upon the "stuff" that the computer is made out of. Lots of people unfamiliar with philosophy of mind imagine that this is what Searle's "causal powers" arguments comes down to -- the particular stuff. Well, it is true that that is one possibility, but Searle, and philosophers in general, do not mean "causal powers" in this context to be taken so simplistically. "Causal powers" could equally well refer to the kind of elaborate symbol grounding mechanisms espoused by Stevan Harnad, "symbol grounding" being another short hand phrase (but a slightly more transparent one) for the abracadabra (or "causal powers") which permits semantics to perfuse the otherwise purely syntactic. >A major problem with the commentaries is that almost no one >really understood all that Searle was saying. There's a lot of it about.... >What he actually said was just too ridiculous. Searle talks about computers in language which to computer scientists is alien, often naive, and sometimes wrong. Nevertheless, his central point, that you can't get semantics out of syntax, is a very important theoretical point for AI and cognitive science, and one which very few of those who like to laugh at his computational solecisms manage to grasp, let alone address. -- Chris Malcolm cam@uk.ac.ed.aipna 031 667 1011 x2550 Department of Artificial Intelligence, Edinburgh University 5 Forrest Hill, Edinburgh, EH1 2QL, UK
thornley@cs.umn.edu (David H. Thornley) (11/20/90)
In article <1990Nov16.171041.14144@ncsuvx.ncsu.edu> fostel@eos.ncsu.edu (Gary Fostel) writes: > > I always wondered why people spent so much time argueing about the Turing > Test when it is so poorly defined. (Or perhaps that's why it can go on and > on). Consider a modification for a chess playing version of the Turing > test. (We could use Chinese Chess if that would help :-) > It is reasonably well-defined. Read Turing's paper. (Please, everyone, read and ponder both one or both of Searle's articles, and Turing's paper, before jumping in. Searle gives me the impression that he is arguing against something of a straw Turing test.) > The only well defined "test", whether it be for chess playing or more > generally, in the Turing test, for intelligence, would be that the machine > could fool EVERYBODY who was observing. This is a very strong test, > much stronger than the usual Turing test I believe, and I'm not sure > many REAL people could convince everyone else that they were intelligent > humans. Perhaps none could! The Turing test is not to convince the observer of intelligence, but to be *indistinguishable* from an intelligent adult human. If you put a standard chess program on line, I can distinguish it immediately by asking about the situation in the Middle East. It seems to me that, if you put two intelligent adults on the lines, instead of one human and one computer, that one of the humans will be identified as the human, and will therefore have passed the test. DHT
thornley@cs.umn.edu (David H. Thornley) (11/20/90)
In article <86001@tut.cis.ohio-state.edu> Todd R Johnson <tj@cis.ohio-state.edu> writes: >.... In other >words, if we are to produce intelligent artifacts we MUST be willing >to accept the fact that we need to create special software AND >hardware. This seems quite reasonable. In fact, I don't see how >anyone can disagree with it. > > ---Todd > If you mean that we need adequate hardware to create intelligence, yes, but nobody has ever claimed that a lawn chair can be made to think with the appropriate algorithms. (OK, I'm probably wrong, but no serious AI researcher has while speaking professionally :-) If you mean that we necessarily need specific hardware not normally found in modern computers, it is easy to disagree with. It seems very likely to me that a powerful enough modern computer, with the adequate software, could "think," in a useful sense of that word. DHT
cw2k+@andrew.cmu.edu (Christopher L. Welles) (11/20/90)
In <3525@aipna.ed.ac.uk>cam@aipna.ed.ac.uk (Chris Malcolm)writes: >In article <8bDqHlK00VsLBAOkxp@andrew.cmu.edu> >cw2k+@andrew.cmu.edu (Christopher L. Welles) writes: > >>Just thought I should point out. Searle did make it clear that "formal >>systems", computers, could be conscious. He emphasized the fact that >>humans were such systems. > >No. He said that machines could understand, that we were no more than >biological machines, but denied that a computer running any program >would be able to do this, on the grounds that in this last case ALL that >is going on is syntactical, and you can't get semantics into a syntactic >process by whatever elaborations of syntactic manipulation. In answer to this, let me pull a quote from Searle's original "The Behavioral and Brain Sciences" article: "Ok, but could a digital computer think?" If by "digital computer" we mean anything at all that has a level of description where it can correctly be described as the instantiation of a computer program, then again the answer is, of course, yes, since we are the instantiations of any number of computer programs, and we can think. "But could something think, understand, and so on solely(in italics) in virtue of being a computer with a right sort of program? Could instantiating a program, the right program of course, by itself be a sufficient condition of understanding?" It is possible to interpret your reply in such a way that it does not conflict with this. However, in doing so, it clearly would not conflict with what I was saying. >>His main thesis went something to the effect of: That there mere >>instantiation of a program could not be in itself, sufficient for for >>consciousness. In this, he is saying that consciousness somehow depends >>upon the "stuff" that the computer is made out of. > >Lots of people unfamiliar with philosophy of mind imagine that this is >what Searle's "causal powers" arguments comes down to -- the particular >stuff. Well, it is true that that is one possibility, but Searle, and >philosophers in general, do not mean "causal powers" in this context to >be taken so simplistically. "Causal powers" could equally well refer to >the kind of elaborate symbol grounding mechanisms espoused by Stevan >Harnad, "symbol grounding" being another short hand phrase (but a >slightly more transparent one) for the abracadabra (or "causal powers") >which permits semantics to perfuse the otherwise purely syntactic. Let me point out that I did not come to conclusion immediately! By means of several arguments, the chinese room test itself is invalid and worthless.( One of the best I've seen of these is in <27320@cs.yale.edu> by mcdermott-drew@cs.yale.edu (Drew McDermott).) However, the issue of intentionality does not need to rest on this in order to have merit. If interpreted as "symbol grounding" it would be a valid point of discussion. This, is something I had taken into account. But, by virtue that it "could equally well refer to" it is ambiguous. Only later is it shown that this is not what he means. The original BBS article consisted of Searle's original paper, commentary on that paper, and Searle's response to the commentary. In that commentary, the "symbol grounding" idea was most closely represented by Fodor's response. His his reasoning was as follows: It is true that merely instantiating a program is not sufficient for consciousness. It must have some sort of causal connection to objects in the real world. This, however, does not mean that those causal connections can not be built. Thus, it does not depend upon the special biological "stuff" the brain is made out of. Searle's response to this clearly shows that he is not referring to "symbol grounding" when he refers to intentionality: "Fodor agrees with my central thesis that instantiating a program is not a sufficient condition of intentionality. He thinks, however, that if we got the right causal links between the formal symbols and things in the world that would be sufficient. Now there is an obvious objection to this variant of the robot reply that I have made several times: the same thought experiment as before applies to this case. That is, no matter what outside causal impacts there are on the formal tokens, these are not by themselves sufficient to give the tokens any intentional content. No matter what caused the tokens, the agent still doesn't understand Chinese. Let the egg foo yung symbol be causally connected to egg foo yung in any way you like, that connection by itself will never enable that agent to interpret the symbol as meaning egg foo yung." In this, Searle returns to the same mistake that he made with the Chinese room itself, and gives up the only valid point his paper could have possibly made. <<<<< Chris >>>>>
fostel@eos.ncsu.edu (Gary Fostel) (11/22/90)
I few posts back, I made the observation that the Turing test was not
a well defined test at all because it does not pin down who's judegement
is to be used in deciding if the test has been passed. No doubt Turing
was expecting it to be conducted by himself, but with him gone, I wonder
who will decide? David Thornley, at the University of Minnesota replied:
It is reasonably well-defined. Read Turing's paper. (Please, everyone,
read and ponder both one or both of Searle's articles, and Turing's paper,
before jumping in. Searle gives me the impression that he is arguing
against something of a straw Turing test.)
I have read Turing original description, and also some of his other musings
on the use of similar tests to see if men and women could tell each other
apart by linguistic means. I spent quite some time discussing it with
AI people and technical philospohers and my conclusion about the weakness
of the test is a strongly held one. Whether or not a machine can pass the
Turing Test ought not to be a function of the judgement of the person who
is trying to apply the test.
To illustrate this point, I used a trivial analogy of a test for chess
playing similar to the Turing test for "intelligence". The point is that
experts in chess, computing, or both, will be far better able to recognize
a computer program playing chess, than will the average person. If we have
a Chess-Test, constructed as the Turing test, then do we accept the judgement
of the experts who can recognize oddities of the computer programs or do we
accept the judegement of an average person? To be more hip, perhaps I should
change this to Chinese Chess and put the chess playing agent in a room. Then
we can all debate Fostel's Chinese Chess Room.
Thornley went on to say:
The Turing test is not to convince the observer of intelligence, but to
be *indistinguishable* from an intelligent adult human. If you put a
standard chess program on line, I can distinguish it immediately by asking
about the situation in the Middle East.
This seems to miss the point I was making. Perhaps Thornley should "read
and ponder" my words before jumping in. The analogy to the chess playing
test is simply a way to amplify the ambiguity present in the Turing test
by using a structurally identical test of properties we understand better.
The same problems exist in the Turing test. Computer programs already exist
that have been confused with intelligent humans, e.g. Weizebaum's Doctor
and the "Paranoid" program (from Stanford?) Observers with different
background (i.e. people on this newsgroup) would not be so easily fooled,
but whose judegement it to be used?
How can the issue of individual judgement be eliminated from the Turing
test? Well, that's the point, it can not. Perhaps one could formulate a
democratic Turing test, and use the consensus of observers to decide if
the agent "passed". Or the Genius Turing test, and have the observer with
the highest IQ make the judgement. Or go for consensus: every observer
must agree. The latter is the only one that make sense. I do not believe
any agent would ever pass that test, if only becuase one or more observers
did not like the personality/politics or lingustic style of the agent.
Remember, we are looking to distinguish a computer from an intelligent human.
How intelligent? Suppose the agent a very stupid and ill-informed human.
They may have no opinion on the current going-on in the mideast, may have a
3rd grade vocabulary and after a while get mad and refuse to co-operate.
Are we going to judge them to NOT be a human? Not intelligent? So far as I
know, we have not yet developed the capability to measure human intelligence
very well; how are we going to specifify a cut-off on the Turing test for
"how intelligent" the agent must be? This is another judegement call for
which we are forced to rely upon the consensus of the observers.
If the agent on the other end of the line is not very intelligent, do we
conclude that they are not human? I hope not. But this suggests that an
easy way to subvert the Turing test is to program a computer to be
surley, uncommunicative, stupid and ill informed. Would it pass? What
would any of this have to say about any of the profound questions of
intelligence, consciousness or minds and machines?
Not a heck of a lot.
It's worth recalling that Turing originally made up the test in the context
of being able to distinguish a person of sex A from a person of sex B
pretending to be a person of sex A. I wonder if Turing himself took it
as seriously as people think. If I imagine him looking down from heaven,
on the current debate, I think he would be chortling in that annoying
high-pitched nasal voice that earned him so many friends. He probably
had the forsight to know that programs like Doctor were feasible and he
was anxious to weaken the then-common dualist belief that humans were the
uniquely chosen vessel of intelligence. Of course, he was an atheist,
so if he IS looking down from heaven, the jokes on him!
----GaryFostel---- Department of Computer Science
North Carolina State University thornley@cs.umn.edu (David H. Thornley) (11/22/90)
In article <1990Nov21.181445.11552@ncsuvx.ncsu.edu> fostel@eos.ncsu.edu (Gary Fostel) writes: > >I few posts back, I made the observation that the Turing test was not >a well defined test at all because it does not pin down who's judegement >is to be used in deciding if the test has been passed. No doubt Turing >was expecting it to be conducted by himself, but with him gone, I wonder >who will decide? David Thornley, at the University of Minnesota replied: > > It is reasonably well-defined. Read Turing's paper. (Please, everyone, > read and ponder both one or both of Searle's articles, and Turing's paper, > before jumping in. Searle gives me the impression that he is arguing > against something of a straw Turing test.) Sorry, that's not the impression I got. This may well have been my mistake; I see lots of misconceptions. > >I have read Turing original description, and also some of his other musings >on the use of similar tests to see if men and women could tell each other >apart by linguistic means. I spent quite some time discussing it with >AI people and technical philospohers and my conclusion about the weakness >of the test is a strongly held one. Whether or not a machine can pass the >Turing Test ought not to be a function of the judgement of the person who >is trying to apply the test. > >To illustrate this point, I used a trivial analogy of a test for chess >playing similar to the Turing test for "intelligence". The point is that >experts in chess, computing, or both, will be far better able to recognize >a computer program playing chess, than will the average person. If we have >a Chess-Test, constructed as the Turing test, then do we accept the judgement >of the experts who can recognize oddities of the computer programs or do we >accept the judegement of an average person? To be more hip, perhaps I should >change this to Chinese Chess and put the chess playing agent in a room. Then >we can all debate Fostel's Chinese Chess Room. > >Thornley went on to say: > > The Turing test is not to convince the observer of intelligence, but to > be *indistinguishable* from an intelligent adult human. If you put a > standard chess program on line, I can distinguish it immediately by asking > about the situation in the Middle East. > >This seems to miss the point I was making. Perhaps Thornley should "read >and ponder" my words before jumping in. The analogy to the chess playing >test is simply a way to amplify the ambiguity present in the Turing test >by using a structurally identical test of properties we understand better. >The same problems exist in the Turing test. Computer programs already exist >that have been confused with intelligent humans, e.g. Weizebaum's Doctor >and the "Paranoid" program (from Stanford?) Observers with different >background (i.e. people on this newsgroup) would not be so easily fooled, >but whose judegement it to be used? > I have not heard that Doctor or Paranoid have passed the Turing test as Turing specified it. Further, let's drop the "Paranoid" test. I can write a version of Catatonic that is guaranteed indistinguishable from a genuine catatonic. >How can the issue of individual judgement be eliminated from the Turing >test? Well, that's the point, it can not. Perhaps one could formulate a >democratic Turing test, and use the consensus of observers to decide if >the agent "passed". Or the Genius Turing test, and have the observer with >the highest IQ make the judgement. Or go for consensus: every observer >must agree. The latter is the only one that make sense. I do not believe >any agent would ever pass that test, if only becuase one or more observers >did not like the personality/politics or lingustic style of the agent. >Remember, we are looking to distinguish a computer from an intelligent human. > >How intelligent? Suppose the agent a very stupid and ill-informed human. >They may have no opinion on the current going-on in the mideast, may have a >3rd grade vocabulary and after a while get mad and refuse to co-operate. > Reread the paper. Look at the example dialogs Turing gives. These aren't the dialogs from an ignorant, inarticulate, human. While I don't remember Turing specifying an intelligent adult, I don't remember him specifying typing ability either. Use a little common sense. >[More on this point. How about a Turing Test for the stupid and > inarticulate?] > >It's worth recalling that Turing originally made up the test in the context >of being able to distinguish a person of sex A from a person of sex B >pretending to be a person of sex A. I wonder if Turing himself took it >as seriously as people think. If I imagine him looking down from heaven, >on the current debate, I think he would be chortling in that annoying >high-pitched nasal voice that earned him so many friends. He probably >had the forsight to know that programs like Doctor were feasible and he >was anxious to weaken the then-common dualist belief that humans were the >uniquely chosen vessel of intelligence. Of course, he was an atheist, >so if he IS looking down from heaven, the jokes on him! > I'm not completely sure about the Imitation Game myself. (This is the version where there is a man and a woman instead of a computer and a human. The man's job is to convince the interrogator that he is a woman; the woman's job is to convince the interrogator that she is a woman. The analogy is exact if you require the interrogator to be a woman.) As far as the definition goes, I will admit that Turing didn't write up the experimental technique like a description in a psychology paper, but the necessary elements are there. If I were to suggest a test for the hypothesis that concrete words are better remembered than abstract, I might say something like "Take some subjects. Give them a list of concrete and abstract nouns combined. Give them recognition and recall tests." I would expect the experimenter to know that subjects should speak the language the words are in. Unfortunately, far too much of the arguments surrounding the Turing test have tried to distort the test in various ways. Did Turing have to explicitly specify that an intelligent and cooperative adult with reasonable typing ability was to be used as the human in the comparison? Do I have to specify that the words should be taken from the subject's native language? So who should judge? I wouldn't trust a psychologist to know how a computer acts, a police officer might consider the psychologist and me to lack practical experience with people's reactions, the list goes on. How about taking various groups of people, such as psychologists, computer scientists, cognitive scientists, police officers, social workers, politicians, any sort of group with some expertise either with people or with computers, and see if any group has an above-chance ability to tell the human from the computer? I'd ask for confidence measures, myself. Personally, I don't think Turing thought it was very important who should judge. I think he considered that any computer that could fool a large number of informed and intelligent observers should be thought of as passing the test. If you want something more specific, he thought that, by 2000, computers with giga-something (byte? digit?) memory would do so well that an average interrogator would identify the human only 70% of the time after five minutes' conversation, and he thought that, by then, people would agree that machines could think. (If you want a specific answer to "who should interrogate," I think this satisfies your requirements.) DHT
cam@aipna.ed.ac.uk (Chris Malcolm) (11/24/90)
In article <wbG=nXC00VsLJNdEVj@andrew.cmu.edu> cw2k+@andrew.cmu.edu (Christopher L. Welles) writes: >In <3525@aipna.ed.ac.uk>cam@aipna.ed.ac.uk (Chris Malcolm)writes: > >>In article <8bDqHlK00VsLBAOkxp@andrew.cmu.edu> >cw2k+@andrew.cmu.edu >(Christopher L. Welles) writes: >> >>>Just thought I should point out. Searle did make it clear that "formal >>>systems", computers, could be conscious. He emphasized the fact that >>>humans were such systems. >>No. He said that machines could understand, that we were no more than >>biological machines, but denied that a computer running any program >>would be able to do this ... >In answer to this, let me pull a quote from Searle's original "The >Behavioral and Brain Sciences" article: > "Ok, but could a digital computer think?" > If by "digital computer" we mean anything at all that has a level >of description where it can correctly be described as the instantiation >of a computer program, then again the answer is, of course, yes, since >we are the instantiations of any number of computer programs, and we can >think. > "But could something think, understand, and so on solely(in >italics) in virtue of being a computer with a right sort of program? >It is possible to interpret your reply in such a way that it does not >conflict with this. However, in doing so, it clearly would not conflict >with what I was saying. Ok, let me quote Searle's reply to the "robot reply" in the original BBS target article and debate: I see no reason in principle why we couldn't give a machine the capacity to understand English or Chinese, since in an important sense our bodies with our brains are precisely such machines. But ... we could not give such a thing to a machine ... [whose] operation ... is defined soley in terms of computational processes over formally defined elements. As I hope is now clear, he is willing to concede the possibility of consciousness to machines "which have a level of description ... described as the instantiation of a computer program", but NOT to anything whose operation is defined SOLEY in those terms. He uses that little word "solely" in both your quotation and mine: it is the important word. The extra magic ingredient required for intentionality (not necessarily consciousness, but it does no damage to the arguments to make that equation here) Searle does often refer to as "causal powers", but I challenge you to find him saying anything anywhere to substantiate your claim that he thinks these caual powers depend - >>>upon the "stuff" that the computer is made out of. This is (IMHO) an extremely common misunderstanding of Searle, based simply on failure of the imagination, i.e., "I can't imagine what else he could have meant". Your later quotation of Searle's rebuttal of Fodor in support of your thesis in fact begs the question: > "Fodor agrees with my central thesis that instantiating a program >is not a sufficient condition of intentionality. He thinks, however, >that if we got the right causal links between the formal symbols and >things in the world that would be sufficient. Now there is an obvious >objection to this variant of the robot reply that I have made several >times: the same thought experiment as before applies to this case. >That is, no matter what outside causal impacts there are on the formal >tokens, these are not by themselves sufficient to give the tokens any >intentional content. You seem to think that Searle's dismissal here of the utility of "outside causal impacts" is equivalent to a dismissal of a functional interpretation of his "causal powers". It isn't. And although you may be correct in saying (I haven't looked back to check) that Fodor's argument is the closest in the original BBS argument to Harnad's symbol grounding point, it is also the case that Harnad would agree with Searle in denying that "the right causal links ... would be sufficient." (though maybe not for the same reasons :-). -- Chris Malcolm cam@uk.ac.ed.aipna 031 667 1011 x2550 Department of Artificial Intelligence, Edinburgh University 5 Forrest Hill, Edinburgh, EH1 2QL, UK
fostel@eos.ncsu.edu (Gary Fostel) (11/28/90)
Is the Turing Test well defined? Thornley, in answer to some of my
objections to serious use of the Turing test was that the person
applying the test needed to:
Use a little common sense.
This is delightful. Is "common sense" related to intelligence? Well,
if so than the key to Thornley's defence of the Turing test might be
that the tester needs to be intelligent. How intelligent? I'm reminded
of the old saw:
Common sense is neither common nor sensible.
My criticism of the Turing test is directed at it's use as an operational
definition of intelligence. It ends up being circular since it can not
be used without slecting a judge to make the decision and the judge must
be intelligent themselves.
A valid SCIENTIFIC test can not rely on the judegement of the experimenter.
That is the point of constructing objective measures and experimental technique
that can be repeated by another researcher. If they do what you do and they
do not get the same result, there is something wrong with the experiment.
I have no trouble imagining that different people will decide differently
about an agent undergoing the Turing test, so the experiment is not valid.
Of course, one answer is to disclaim any intent to be scientific about it
and go back to argueing about what it all means in a subjective sense. It
is pointless to seriously think about constructing a test for the presense
of a thing that may not be a well defined thing at all. My own view of
what intelligence is, will be different from yours and that pretty much
scotches any serious scientific work until someone comes up with a more
testible property than "intelligence". There is probably some way
to convert the Turing test into a useable measurement instrument, but I
hope anyone doing that would not claim they were measuring or detecting
"intelligence".
----GaryFostel---- Department of Computer Science
North Carolina State University thornley@cs.umn.edu (David H. Thornley) (12/01/90)
In article <1990Nov27.231501.1621@ncsuvx.ncsu.edu> fostel@eos.ncsu.edu (Gary Fostel) writes: > >Is the Turing Test well defined? Thornley, in answer to some of my >objections to serious use of the Turing test was that the person >applying the test needed to: > > Use a little common sense. > >This is delightful. Is "common sense" related to intelligence? Well, >if so than the key to Thornley's defence of the Turing test might be >that the tester needs to be intelligent. How intelligent? I'm reminded >of the old saw: > > Common sense is neither common nor sensible. > >My criticism of the Turing test is directed at it's use as an operational >definition of intelligence. It ends up being circular since it can not >be used without slecting a judge to make the decision and the judge must >be intelligent themselves. > >A valid SCIENTIFIC test can not rely on the judegement of the experimenter. True. So what's wrong with Turing's suggestion of "an average interrogator will not have more than 70 percent chance of making the right identification after five minutes of questioning"? This is a specific proposal, if you will grant a little leeway on the word "average." (In psychology departments, it seems to be the average introductory psych student.) When I read a paper claiming that a system has passed the Turing test, I want to read how the interrogators were selected, what sort of relevant backgrounds they had, and a statistical breakdown of success rates. I also want to see confidence ratings, and various other assorted details. I then intend to study the article carefully. I certainly would not take a claim seriously without such background. In particular, if someone were to claim that their husband/wife could not tell the difference, and that is how the machine passed the test, I would not be impressed. In the meantime, please feel free to propose any selection process you please for the interrogators (I insist on multiple interrogators). I don't think you will, in good faith, come up with one so weird that I will refuse to consider it a Turing test. However, I claim that it makes sense to speak of the test in general terms, with the understanding that we will nail it down when necessary. Describing something as "passing the Turing test" is something like referring to "the capacity of short-term memory" in that we don't quite define it as we say it, but we can use it in the lab just fine. DHT