 Philosophy Area |
The Great Philosophers See the Alan Turing Home Page for a guide to this website. |
Part 8 of Turing: a natural philosopher (1997)
The Turing Test
Turing moved to Manchester University, where Newman, professor of pure mathematics there since 1945, had secured for him a first full-time academic post. Turing had a difficult position as software writer for the pioneer computer, the first such in the world, that the electronic engineer F. C. Williams and his team had built after Newman gave them the principle. Press reports of the machine had already used the terminology of 'brains', which Turing's own comments in 1949 did nothing to discourage. Jefferson, a Manchester brain surgeon, attempted to dispel such talk in a 1949 lecture. Michael Polanyi, chemist turned Christian philosopher of science at Manchester, was another intellectual opponent (with whom, however, Turing was on friendly personal terms.) It was probably Polanyi who suggested that Turing should present his views as a paper, which appeared as Computing machinery and intelligence in 1950. [11]Turing addressed the problem of writing for a non-mathematical readership with typical sang-froid, ignoring all conventional cultural barriers. Notably without any serious citations in philosophical or psychological literature, it stands intransigent in style as well as content.
The paper is now famous for the 'imitation game', as described below, and now often called the Turing Test. But the most solid aspect of the paper is Turing's setting out of the model of the discrete state machine, which is the Turing machine of 1936, but more clearly thought of as being physically embodied. A careful paragraph explains first why computing machinery is discrete:
Digital computers... may be classified amongst the 'discrete state machines'. These are the machines which move by sudden jumps or clicks from one quite definite state to another. These states are sufficiently different for the possibility of confusion between them to be ignored. Strictly speaking there are no such machines. Everything really moves continuously. But there are many kinds of machine which can profitably be thought of as being discrete state machines. For instance in considering the swiches for a lighting system it is a convenient fiction that each switch must be definitely on or definitely off. There must be intermediate positions, but for most purposes we can forget about them.
The special property of digital computers, that they can mimic any discrete state machine, is described by saying that they are universal machines.
Turing's argument is simply that the brain should also be considered as a discrete state machine. In his classic statement, made in a 1952 radio broadcast [12]:
We are not interested in the fact that the brain has the consistency of cold porridge. We don't want to say 'This machine's quite hard, so it isn't a brain, so it can't think.'
The physical greyness or soft sponginess of the brain is irrelevant, and so is the mode of operation of the nerves:
Importance is often attached to the fact that modern digital computers are electrical, and that the nervous system also is electrical... Of course electricity usually comes in where fast signalling is concerned, so that it is not surprising that we find it in both these connections. In the nervous system chemical phenomena are at least as important as electrical. In certain computers the storage system is mainly acoustic. The feature of using electricity is thus seen to be only a very superficial similarity. If we wish to find such similarities [i.e. significant similarities between brain and computer] we should look rather for mathematical analogies of function.
Turing's claim is that the only features of the brain relevant to thinking or intelligence are those which fall within the discrete-state-machine level of description. The particular physical embodiment is irrelevant. Not quite made explicit, but implicit in every statement, is that the operation of a discrete state machine is computable. We now see the definitive extension to the argument presented in 1936, the effect of the change of thought I conjecture for 1941. The post-war Turing claims that Turing machines can mimic the effect of any activity of the mind, not only a mind engaged upon a 'definite method.'
Turing's term 'discrete state machine' is a judicious choice. He avoids expressions such as 'logical structure' which might carry the false connotation of common parlance: logical as opposed to illogical, informal or emotional thought. There is no such dichotomy in Turing's analysis, and indeed no distinction between conscious and unconscious. Turing is clear that discrete state machines include machines with learning or self-organizing ability, and makes a point of the fact these still fall within the scope of the computable. Turing draws attention to the apparent conflict with the definition of Turing machines having fixed tables of behaviour, but sketches a proof that self-modifying machines are still in fact defined by an unchanged instruction set, ending:
The explanation of the paradox is that the rules which get changed in the learning process are of a rather less pretentious kind, claiming only an ephemeral validity. The reader may draw a parallel with the Constitution of the United States.
If Turing's thesis about the function of the brain is accepted, then from a materialist point of view, the argument is almost complete. The behaviour of a discrete state machine can then, at least in principle, be written down in a table. Hence every feature of the brain relevant to thought can be captured by a table of behaviour, and so emulated by a computer. The only question that might remain is that of whether the speed and spatial dimensions of the brain, and the character of its physical interface with the world, are also essential to its function.
The imitation game
However, the rest of the paper, bringing in the definition of the Turing Test, does much to illustrate the idea of a brain and its function as a physical object whose properties can be examined like any other, and to suggest constructive methods by which intelligent machinery could be engineered. To do this, Turing dramatizes the operational viewpoint. Instead of considering the question 'Can machines think?' Turing explains, 'I shall replace the question by another, which is closely related to it and is expressed in relatively unambiguous words':The new form of the problem can be described in terms of a game which we call the 'imitation game'. It is played with three people, a man (A), a woman (B), and an interrogator (C) who may be of either sex. The interrogator stays in a room apart from the other two. The object of the game is to determine which of the other two is the man and which is the woman.If Turing's introduction to the problem of creative mind through a party game with camp innuendo was intended to offend the pretensions of Arts-educated intellectuals, it probably succeeded. Unfortunately Turing also succeeded in creating disastrous confusion. Although it was intended to clarify the picture of the brain being tested like any other physical object, for many readers the purpose of the imitation game is totally obscured, in fact turned on its head, by careless syntax:
We now ask the question, 'What will happen when a machine takes the part of A in this game?' Will the interrogator decide wrongly as often when the game is played like this as he does when the game is played between a man and a woman?I have no doubt that 'the game is played like this' means, 'the game is played between a human being and a computer pretending to be human'. But many are now the books, articles, lectures, and webpages which assert that in the Turing Test the computer must take the part of a man who is imitating a woman. This is indeed the literal meaning of the words 'a machine takes the part of A', but such an interpretation is contradicted by the sample interrogation:
Q: Please write me a sonnet on the subject of the Forth bridge.These answers establish no impression of gender; they are meant to establish human intelligence (including — a subtle point — the incorrect addition.) The point of the game is this: if a machine cannot be distinguished from a human being under these conditions then we must credit it with human intelligence.
A: Count me out on this one. I never could write poetry.
Q: Add 34957 to 70764.
A: (Pause about 30 seconds and then give as answer) 105621.
Q: Do you play chess?
A: Yes.
Q: I have K at my K1, and no other pieces. You have only K at K6 and R at R1. It is your move, What do you play?
A: (After a pause of 15 seconds) R-R8 mate.
A deeper problem is that Turing's gender-guessing analogy detracts from his own argument. In the gender game, successful fooling of the interrogator proves nothing about the reality behind the screen. In contrast, Turing wants to argue that the successful imitation of intelligence is intelligence. Equivalently, Turing defines the subject matter of intelligence as that which can be wholly communicated via the teleprinter link, consistent with his thesis that the brain is relevant only qua discrete state machine. Discrete symbols over the teleprinter link can faithfully represent all the inputs and outputs to and from a discrete state machine. As Turing puts it:
The new problem has the advantage of drawing a fairly sharp line between the physical and intellectual capacities of a man.The setting of teleprinter communication is intended to separate intelligence from other faculties of the human being.
We do not wish to penalise the machine for an inability to shine in beauty competitions, nor to penalise a man for losing in a race against an aeroplane... The conditions of our game make these disabilities irrelevant.The conditions are intended to make qualities such as gender irrelevant; and from the point of view of clarity it is unfortunate that his iconoclastic introduction gives the opposite impression.
But if Turing's gender-game is misunderstood, he certainly courted such confusion. He painted the pages of this journey into cyberspace with the awkward eroticism and encyclopaedic curiosity of his personality. Modern cultural critics have jumped with delight to psycho-analyse its surprises. The intellectual text is the austere statement of the capacity of the discrete state machine for disembodied intelligence; the subtext is full of provocative references to his own person, as if putting his own flesh-and-blood intelligence on trial.
It might be said that the 'imitation' intrinsic to the Turing test is also a distraction from the core of the argument. Subjecting 'imitation' to analysis raises questions, undiscussed in Turing's paper, such as why an intelligent machine should be expected to play a dishonest game; in my view this is overburdening Turing's illustration and missing the main point. The real claim, as I have emphasised above, is that the brain's function is that of a discrete state Turing machine, and therefore can be performed by a computer. The colour and drama in Turing's writing is secondary: it is intended to invite a wide variety of readers into constructive reflection upon this conclusion, which runs contrary to all intuition but is not at all easy to refute.
[11] A. M. Turing. 'Computing machinery and intelligence'. Mind 51 (1950) pp. 433-60.
[12] Transcript in the Turing Archive, King's College, Cambridge. (See the Bibliography on this site.)
© 1997, Andrew Hodges.
CONTINUE to part 9
Book index | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12
| Quick Links: |
|
|
|
|
|
|
|
| |||||