The Nature Of The Machine
AI and Wittgenstein: a possible dialogue.
Introduction
I present the thesis that follows in the form of a dialogue. Although it may become obvious by the end of the paper what the setting for this dialogue is, I feel that I must explain it here since it will certainly be at least slightly confusing until it is understood. What takes place is a conversation between two students of Wittgenstein, sometime in a possible future. In this future the tenants of education have been defined in such a way that it is wrong, and indeed illegal, to use any sources outside those used in any particular course. Hence, there is apprehension, particularly on the part of speaker “B”, to refer to such people as Searle and Locke. If they were “found out” to be doing so the consequences could be severe.
I should also mention that I don’t present this “future”, or form of dialogue, for any particular reason; it is just a setting for the dialogue that “happened” as I began to write. Both the 1984 type of control and intolerance, and the impersonal use of “A” and “B” to describe the speakers (although these need not be their real names), are not an indication of anything specific. It is the content of the dialogue that should be examined.
| JB: | This essay, as assigned in class, was based on Wittgenstein’s Philosophical Investigations 3rd ed., Basil Blackwell Ltd., translated by G.E.M. Anscombe. |
| A: | Hey, guess what I found! | ||||||
| B: | What’s that? | ||||||
| A: | This old philosophical story dealing with Searle’s views on strong AI and how he misinterpreted the problem. | ||||||
| B: | Well, ok. But why are you so excited about it? We’ve been discussing Wittgenstein’s theories in class, and although AI is one of the subjects we’ve looked at, you know as well as I do that Searle’s not part of the course material. | ||||||
| A: | I wasn’t that enthusiastic about it either until I started reading it, and you’d be surprised just how inspirational it was for me…You know I think that this policy of ours to specialize in one field to the exclusion of all else is misplaced. | ||||||
| B: | Shhh! Do you know what you’re saying? What if somebody should hear? | ||||||
| A: | Look, there’s nobody around at present. Besides if we’re careful about it no-one will ever find out. And I do think that you’ll find it quite interesting. | ||||||
| B: | Umm…ok. Let’s hear it. | ||||||
| A: | Right. It’s from an essay written for an undergraduate course in April 1990, by a student named Jason Bassford. It’s called, “A Scientist And His Android: A critique of Searle’s views on strong AI“. Wait just a minute while I remember – you realize that I wouldn’t have been foolish enough to bring the actual essay? | ||||||
| B: | Oh, of course! | ||||||
| A: | It took me a while to memorize it. Now, what was that opening line? Oh, yes. | ||||||
| A: | And it ends there. What do you think? | ||||||
| B: | Ah, I don’t know. It was an interesting story. | ||||||
| A: | Oh, Lord. Look, let’s assume for the sake of argument that Bassford’s analysis of Searle’s arguments against AI is correct. | ||||||
| B: | Ok. | ||||||
| A: | Now, do you think that the same arguments can be applied to what Wittgenstein would say against AI? | ||||||
| B: | I get it…Obviously not! The two philosophies are quite incompatible. | ||||||
| A: | In the main I might agree with you. But consider “Rex’s” refutation of the Chinese room analogy. It seems to me that if it is true that Searle was begging the question concerning the non-Chinese speaker in the room, and hence the fact that the speaker could not really have an understanding of Chinese, that the same could be said of Wittgenstein himself. | ||||||
| B: | I don’t follow you. | ||||||
| A: | Alright then, look at proposition #283 in which he says that, “Only of what behaves like a human being can one say that it has pains.” Further, in 360, he says that a machine cannot think, that, “We only say of a human being and what is like one that it thinks.” | ||||||
| B: | Well, of course. I wouldn’t go around ascribing intelligence to a rock or to a tree. There has to be some standard by which we can say that something is capable of thought. We all know that we think, that human beings are capable of thought. It stands to reason that anything which is like a human being, as like enough as to make us believe that it is capable of thought, is capable of thought. | ||||||
| A: | And why does there have to be this standard, this idea of being a person or person-like? Are you saying that because we have no evidence of the thinking of anything which doesn’t give us such evidence, that therefore it doesn’t think? That seems a little close-minded to me. After all, before we had any evidence for the spherical shape of the planet it was believed that it was flat. | ||||||
| B: | Don’t tell me you believe that a rock thinks! | ||||||
| A: | No, I don’t believe that. However, what does what I believe have to do with it? I could very well believe that it doesn’t think, as I do, only to discover one day that it is communicating with me in some way, informing me that it does in fact think. I could believe that the world was flat -and I would be wrong. | ||||||
| B: | You could believe any number of things and be wrong or right. But so what? This is an absurd argument. We can’t get bogged down in such fantasy situations as a way of proving something. I could think of any number of counter-examples. You might well believe the rock to think – and be wrong! As you undoubtedly would be. No, there has to be some point at which we cannot speculate any further. Because otherwise we would just lose ourselves in nonsense. | ||||||
| A: | Alright then, forget that. Let’s say that there’s a need to say that only those things that are persons, or at least person-like, are capable of thought as Wittgenstein says. Just what standards are you going to apply to a definition of person-hood? | ||||||
| B: | None at all! | ||||||
| A: | But surely you’re going to have to tell me what you mean by the word “person” if I’m to follow any arguments that you come up with which use that term. | ||||||
| B: | No, I don’t think so. Don’t you know what I mean when I say that so-and-so is a person? | ||||||
| A: | That depends on how you wish me to interpret that question. | ||||||
| B: | Oh, come on! Your son asks you one day as you’re walking down the street if a mailbox is a person. You’re going to tell him that it isn’t, aren’t you? | ||||||
| A: | Yes I am. Similarly I would just as well tell him that Mrs. Canfield across the street from us is a person. But what’s your point? | ||||||
| B: | Just this. You know perfectly well if something is or is not a person. You have no problem at all in using the word. And trying to define the word in a way that is absolute, in a way that I can see that you want to do, so that you can do philosophy “properly” as you might put it – that is quite hopeless. As Wittgenstein points out, it is quite impossible to give a complete description of the aroma of coffee; all that can be done is to say that it has this kind of smell, that it is like this or that. Such things are of a different sort than might be chemical or mathematical equations where it is quite simple to say that, “Two plus two is equal to four.” These other things are things that are defined. When it comes to coffee aroma or person-hood all we can do is offer indications as to what we mean. And after all, in the end, these are quite enough. Our meaning is shown by the use to which we put the words. | ||||||
| A: | Yes, alright then. But you still haven’t addressed the real question. What you have said amounts to one thing; that a person is a person because we call them a person. This is like saying that the Earth is flat because the Earth is flat. In some cases, like the aforementioned mathematics where two plus two is defined as four, such an axiom might be acceptable but in the cases that we want to consider, it certainly is not. We are not dealing with the same thing at all. I have no need to prove to anyone that two plus two is four, that is a necessary consequence of the meanings of the terms “two”, “plus”, and “four”. But I do need to offer proof for the roundness (or flatness) of the Earth, just as I do need to offer proof for the person-hood of something. Otherwise I could say anything that I wanted. | ||||||
| B: | But you don’t say anything that you want! There is no such thing as language suddenly changing its usage at random, being different from one person to the next from one time to another. We do know what “person” means because we use it in a certain way. And this is something that is proven by its use. | ||||||
| A: |
Alright, if that’s the case then you tell me why we are having so much confusion over the question of AI. Why is it that I would wish to say that “Rex” is a person and you would wish to say that he is not? After all we both apparently understand the meaning of the word “person”, we both use it in the same way, and I think that Rex is a person while, it seems, you do not. |
||||||
| B: | Well, since it seems that we do disagree I can only say that we don’t use “person” in same way. If we did then there would be no room for disagreement. If I took the meaning of the word in the same way that you did then there we would naturally agree. But it seems that we are, as it were, playing at different language games. | ||||||
| A: | Yes…and? Are you telling me that that’s it? Are you trying to say that whenever there is disagreement that there is no point in going any further? This cannot be the case because many people, many times the world over, have had arguments and been able to settle them. It is by one person…excuse me… | ||||||
| B: | You see! | ||||||
| A: | …showing another that they were wrong, that they are indeed playing the same language game; only that they have simply misinterpreted the rules. After all in chess, if I move my knight in an illegal way that would result in my king’s capture, it isn’t going to be said that I wasn’t playing chess. No, what is the case is that I simply made a mistake. Most likely I would make an apologetic sound once this was pointed out, take the move back, and continue in the proper fashion. I’m afraid that you can’t really mean that all disagreement stems from different language games, assuming of course that you are going to go along with Wittgenstein. The claim that a computer is capable of thought, is simply for him something that is nonsense. It isn’t that such a claim is a move in a different language game but that it isn’t a move in any language game at all. | ||||||
| B: | Yeah, alright. So what’s your point then? | ||||||
| A: | My point is that, while Wittgenstein would say that the idea of AI is nonsense, I hold to the view that it is quite the contrary. I believe that the AI view, that a computer is a person, or something that is person-like and to which we can ascribe person-attributes, even if we don’t consider it to be a person in the human-is-equal-to-person sense of the word, that the AI view really is a different language game as you originally suggested. There are, after all, enough philosophers around at present, enough talk about AI within a large section of the community, to make it a real language game that is used…(Yes, I know what you’re going to say. You’re going to say that just because it is commonly believed that something is true that that doesn’t make it true. In the first place I’m not arguing that AI does exist. Like Rex I’m only trying to show that it is a possibility. And in the second place I don’t want to get side-tracked down that line of reasoning at present. Alright?)…To continue, the reason that AI and anti-AI views can never agree is that they have no basic common ground upon which to do so. It is, after all, impossible to reconcile the game of war with the game of peace, regardless of what those in power might wish us to believe. | ||||||
| B: | Don’t say that! | ||||||
| A: | Stop being so overcautious. As I said before there’s nobody else here right now. Do you want me to proceed or not? | ||||||
| B: | Yes, yes. | ||||||
| A: | Then just listen. Assuming that I’m right about these differing language games, and I freely admit that there is considerable room for argument, then I maintain that Wittgenstein is begging the question in a fashion similar to the way in which Rex believed that Searle was. | ||||||
| B: | But there is no question about an English or non-English speaking person here. In Bassford’s essay he brings up the fact that in order for us to follow a rule in a way that a computer cannot there must be something in us, something apart from the rule, which does the following. He says that there must be something which understands the rule. But this isn’t true! There is only “English being spoken”. | ||||||
| A: | And it would seem then that you’ve just stated that there is nothing different between a person and a computer. If there is nothing different in a person’s manipulation of symbols than that of a computer’s, if there is nothing in a person which understands, then both can be discussed in the same sense. It would seem, my friend, that you have unwittingly proven the very point that you wished to deny. | ||||||
| B: | Uh…no, wait! That’s not right. Remember the debate in 304: “‘But you will surely admit that there is a difference between pain-behaviour accompanied by pain and pain-behaviour without any pain?’ – Admit it? What greater difference could there be? – ‘And yet you again and again reach the conclusion that the sensation itself is a nothing.’ – Not at all. It is not a something, but not a nothing either! The conclusion was only that a nothing would serve just as well as a something about which nothing could be said. We have only rejected the grammar which tries to force itself on us here.” It isn’t the case that there is nothing which understands the rule. It is simply the fact that there isn’t something (a something that we can talk about) which does this…And you must have known that! | ||||||
| A: | Yes, I admit it. I concede defeat on this particular point. However, what I do not concede is the question begging attitude of Wittgenstein that I introduced earlier. If it is true that there is some kind of “non-entity” which gives rise to understanding, a non-entity without which understanding is impossible, then even so there is still just as much an instance of question begging now as there was before – in fact more so. | ||||||
| B: | And just what makes you say that? | ||||||
| A: | The simple expedient of Wittgenstein identifying this non-entity with humans – persons – and not identifying it with computers – non-persons. For what reason can he offer in doing this? Yes, perhaps it is true that the word “person” is used in such a way that we can attribute this non-entity to humans and not to computers but, as I believe I have made clear, there may also be a sufficient number of people who attribute this non-entity to computers as well. If this is the case then we are really back to where we started: the simple fact that Wittgenstein says that persons are persons because they are and computers aren’t because they aren’t, in spite of the fact that the usage of the word, the criterion for such a claim, may be different in some cases than he says it always is. Where Rex would question the validity of assigning a “something” which understands to humans and not to computers, I question the validity of so assigning, and not assigning, this “non-entity” of person-hood to those same things. | ||||||
| B: | But…then we’re just back to a position in which we can claim person-hood to anything at all. You’ve just returned to the possibility of assigning consciousness to the very rock that you said wasn’t conscious! | ||||||
| A: | No, I haven’t. I’m not about to dispense with the idea of meaning as use; as a matter of fact the force of my above argument rests very much on such an idea – an idea that makes quite a great deal of sense to me. No, the rock is not a person in any sense of the word that I’m aware of…except maybe in some Eastern interpretations – but I’m certainly not going to get into that. I’m sorry, I digress. The rock is not a person. While there may be disagreement over the status of the computer (it may be argued that the AI viewpoint is part of a certain language game after all) it cannot be said that the rock-as-person viewpoint is part of any language game. There is no disagreement over this matter at all. | ||||||
| B: | But, if I understand you correctly, shouldn’t there be disagreement? I mean if the non-entity is something that is unverifiable as you claim, then how could you rule out the rock after all? | ||||||
| A: | I shall tell you how, and in so doing I wish to raise another point of criticism in regards to Wittgenstein. He places a great deal of emphasis on the denial of any kind of “private object”. By this I mean an object that can only be understood by the subject who is aware of the object; a pain, emotion, or thought for example. If it were the case that a private object existed then it must be possible for the subject perceiving that object to describe what it is that they perceive. Further, since it is impossible for there to be a private language (a point that I am quite willing to accept), then it must be the case that the description of this private object takes place within a public language. But if this is possible then it would no longer be the case that the object was private since, by definition, if it is described in a public language it must be something that can be understood by people other than the “private subject”. | ||||||
| B: | Well, this is quite evident. I can see no way for you to break this chain of reasoning. | ||||||
| A: | Oh, I don’t intend to break it, but rather the bedrock to which it is attached. Can you recite the famous “beetle in the box” example for me? | ||||||
| B: |
293? I would be delighted:
And that’s it. So what do you have to say about it?…Well? |
||||||
| A: | Oh, I’m sorry it’s just that I’ve realized something that I should have told you before. | ||||||
| B: | What’s that? | ||||||
| A: | The line of argument that I’m going to follow is not an idea that I can claim to have thought of on my own. | ||||||
| B: | Oh, yeah? | ||||||
| A: | I have to admit that I have gone to yet another philosopher outside the scope of our course. His name is Don Locke and he presented the following criticism of Wittgenstein in a little book called, “MYSELF AND OTHERS A Study in Our Knowledge of Minds“. | ||||||
| B: | I see… | ||||||
| A: | I know how sensitive you are to sticking to the school’s guidelines for proper education, as well as staying in line with our present government’s policies…Is it alright with you if I continue anyway? | ||||||
| B: | I can almost begin to see why the policies of yesterday were better…I don’t think we could have had such an enlightening conversation without “breaking the rules” the way we are now. Yes, I’m too fascinated to stop. | ||||||
| A: | Alright, I won’t speak of it again. In any case, what Locke proposes, a position that I find myself sympathizing with, is that Wittgenstein misconstrues the nature of a private object, such as the beetle that you just mentioned. | ||||||
| B: | I don’t understand. Are you saying that we can use the word “beetle” to describe something that may not only be constantly changing but which might actually not exist, and all of this at the same time that it might be something different for different people? This must be wrong! | ||||||
| A: |
Indeed it is. The mistake that Wittgenstein makes is in the way that he says we use the word “beetle”. He sees it as referring to something, when the function that it really serves is to mean something. While there may be no consistent reference (and there can be no mistake that there is something in the box, else obviously the person who looks in their box and sees nothing would surely be able to say, acquainted as they are with the concept of “nothing”, that there was nothing in their box, whatever the supposed thing that “beetle” was supposed to refer to), while there may be no consistent reference, there is nonetheless a consistent use of the word. This use lies is the fact that “beetle” means whatever it is that is in the subject’s box. If I see an “x” and you see a “y”, and there is no way to tell that what I see is any different from what you see, it does not mean that we cannot communicate about the thing-in-the-box. As Locke says on page 106 of his book:
And similarly, in response to a criticism that I suspect you are just about to voice, the same can be applied to the individual. Just because the pain that that person might feel may be different from a pain they might have felt in the past does not mean that they are in error by ascribing the word “pain” to it. In all cases of the use of that word, what it means is, again, “‘whatever it is that we feel in such and such circumstances.'” Locke sums this all up quite nicely on page 109 when he says that Wittgenstein
|
||||||
| B: | Well, I must say that I’m not quite sure what to make of that! | ||||||
| A: |
It might perhaps be useful if you were to think on some of Wittgenstein’s own words; words which, I think, support Locke’s thesis. Remember 380-381:
It would certainly seem as if he is pointing towards a “meaning/use” definition of the word red rather than a “reference” definition. He doesn’t want to say that we necessarily see the same thing at all when we look at this, but rather that we use the English word “red” in the way that we have been taught to; namely that whatever it might be that we see when we look at this, we shall nonetheless call it “red” and understand it to be “red”. How then can he object to private objects at the same time? If we have to believe one of these over the other it seems more natural to me to believe the former as it relates more directly to his general thesis of language and meaning as use. |
||||||
| B: | But surely something like “red” is of a different type than something like “pain”. The first is something that everybody can have access to, it is “out there” in the objective world, it can be empirically verified. Whereas “pain” is nothing of the sort. All that can be said about pain is that “I” have it and only “I” know that “I” have it. | ||||||
| A: | They are really not of any different type at all, if you consider the context in which they are placed. We are not concerned with the light spectrum of the thing that we call red, which is indeed something empirical and verifiable – it can be measured as x on some scale. What we are concerned with is the internal, subjective perception that we have when we look at that thing (and when we feel pain and so on). There is no denying that that thing has a reading of x on some scale, but it can be denied that each of us perceives it in the same way. Wittgenstein talks of recognizing something as red, not of it actually being red (whatever that might really mean). The point that he makes is that our subjective experiences become objective when we use language in a way that makes our experiences communicable, and hence meaningful, to others, if not meaningful in and of themselves. It is in this important sense that “red” and “pain” are very much things of the same type. | ||||||
| B: | Hmmm…But what about my original point? How is it that you can say that a rock does not possess this “non-entity” which is unverifiable? Surely if it is not verifiable than you cannot be able to tell just what possesses it. | ||||||
| A: | Ah, yes. The answer is contained in the whole “meaning as use” issue. In presenting an argument for the existence of private objects, as things about which a meaning if not a reference can be ascribed, what I did was to show that a “beetle”, for example, is a “beetle” by virtue of its being a description of something (whatever that might be) that is in a box. Similarly we cannot say that a “pain” is something that we feel whenever we are hungry and eat something. No, we might rather call this feeling contentment, pleasure, etc. When we say “pain” we mean whatever it is that we feel when we say, “Ouch!,” or jump up and down holding some part of our body, or say, “I’m in pain,” or take an aspirin, etc. It is by virtue of the context of the word “pain” that we are able to judge what “pain” is and whether or not someone is in pain. | ||||||
| B: | And what are we to make of the fact that you could very well be in pain and not exhibit any pain-behaviour? Or that you could not be in pain while still saying that you are? | ||||||
| A: | It is certainly true that you know what pain is, as you have experienced something that was associated with your own past pain-behaviour. You have come to identify a certain sensation as that which belongs to the idea “pain”. You may very well be in pain and not exhibit any pain-behaviour. Nonetheless we would not wish to say that this pain-behaviour is something that you instinctively repress, indeed it is almost certain that you would have to be making a conscious effort to not express pain-behaviour after you had been shot, stabbed, broken a limb, or whatever. Knowing that you naturally ought to be expressing pain-behaviour, and yet not doing so, is itself indicative of the fact that you know you are in pain – as is the other simple fact that you are experiencing a sensation that you recognize from past experience as “pain”. Similarly, it wouldn’t be natural to express pain-behaviour when you feel no pain. You must recognize the absence of pain for the same reasons as you do its presence. | ||||||
| B: | And the rock? | ||||||
| A: | The rock in no way behaves as if it is intelligent. It doesn’t communicate anything in any way at all. It is not the type of thing that we might even be tempted to describe as intelligent. Whereas in contrast, the computer does exhibit behaviour that we might be tempted to describe as intelligent. While we may not be able to tell if something does have that elusive non-entity to which understanding is attached, we can certainly tell if something does not. If it exhibits no behavioral signs whatever that might tempt us to say that it possesses such a thing, then we can state that it does not possess it. | ||||||
| B: | Then you are a behaviourist? | ||||||
| A: | Not at all. Nor is Wittgenstein, as I’m sure you’re aware. I do not say that all there is is behaviour. What I do say is that we may only judge things by their behaviour, and give things meaning in the context of that behaviour. Wittgenstein says in 307 that if he does, “…speak of a fiction [that is, mental processes], then it is of a grammatical fiction.” | ||||||
| B: | And what of, in the extreme case, the person in a coma? Are you going to claim that because they exhibit none of the signs of being a person that they are not a person? That is not a position that I would want to defend. | ||||||
| A: | Nor would I, since I sense that you are using the word “person” in a different sense than we have been using it so far. By “person”, what seems to have been meant up to now is someone who can think. In this sense the coma patient exhibits no signs of being a person and indeed, for as long as they continue to exhibit none of these signs, could not be said to be a person. However, this is not the sense in which you just now used the word “person”. You wanted to indicate by the word that the coma-patient is not to be treated as a thing, not to be done with as anyone might will on a whim. Is this not correct? | ||||||
| B: | Yes… | ||||||
| A: | And in that sense it might be claimed that “person” means someone who has shown thinking-behaviour before and might again, or someone who is exhibiting life-behaviour, or whatever. In any case the point is that the (behaviour) criteria for “person” as it means “thinking being (at present)”, is not the same as for “person” as it means “being who should be treated with moral consideration”. Although that is not to say of course, that one does not necessarily influence the other in some way. | ||||||
| B: | And where does all of this leave us? | ||||||
| A: | It leaves us with the conclusion that there are arguments in favour of AI, an opinion that Wittgenstein did not hold; it shows that Wittgenstein may have been as guilty as Searle in begging the question when it comes to humans being persons (things that think) and computers not being persons; it also indicates a possible inconsistency between his belief in the meaning of words being dependent on the way that we use them and in the impossibility of private objects. | ||||||
| B: | I see. You will excuse me if I still reject the notion that AI is possible. While you have offered intriguing arguments against Wittgenstein I don’t think that you are right. The next time we meet I will have had a chance to think things over and I expect by then to have something more to say on the matter! | ||||||
| A: | Of course. It is to be expected, even hoped for, that we disagree. Where else but through thesis and antithesis can synthesis be created? I’m also perfectly willing to admit that my arguments are not unassailable. I look forward to our next meeting as much as you do. |