He imagined a game in which an interrogator would have to decide, on the basis of written replies alone, which of two people in another room was a man and which a woman. The man was to deceive the interrogator, and the woman to convince the interrogator, so they would alike be making claims such as ‘I am the woman, don’t listen to him!’ Although pleasantly recalling the secret messages that might be passed in his conversations with Robin and Nick Furbank, this was in fact a red herring, and one of the few passages of the paper that was not expressed with perfect lucidity. The whole point of this game was that a successful imitation of a woman’s responses by a man would not prove anything. Gender depended on facts which were not reducible to sequences of symbols. In contrast, he wished to argue that such an imitation principle did apply to ‘thinking’ or ‘intelligence’. If a computer, on the basis of its written replies to questions, could not be distinguished from a human respondent, then ‘fair play’ would oblige one to say that it must be ‘thinking’.
This being a philosophical paper, he produced an argument in favour of adopting the imitation principle as a criterion. This was that there was no way of telling that other people were ‘thinking’ or ‘conscious’ except by a process of comparison with oneself, and he saw no reason to treat computers any differently.*
The Mind article largely took over what he had said in his NPL report, which had not, of course, been published. There were, however, some new developments, not all very serious. One was the joke of a proud atheist who refused to be the Responsible Scientist expected by Downside Abbey. He gave a tongue-in-cheek demolition of what he called the ‘Theological Objection’ to the idea of machines thinking, which concluded that thinking might indeed be the prerogative of an immortal soul, but then there was nothing to stop God from bestowing one upon a machine. More ambiguous in tone was a reply to an objection ‘from Extra-Sensory Perception’.
He wrote that
These disturbing phenomena seem to deny all our usual scientific ideas. How we should like to discredit them! Unfortunately the statistical evidence, at least for telepathy, is overwhelming. It is very difficult to rearrange one’s ideas so as to fit these new facts in. Once one has accepted them it does not seem a very big step to believe in ghosts and bogies. The idea that our bodies move simply according to the known laws of physics, together with some others not yet discovered but somewhat similar, would be the first to go.
Readers might well have wondered whether he really believed the evidence to be ‘overwhelming’, or whether this was a rather arch joke. In fact he was certainly impressed at the time by J. B. Rhine’s claims to have experimental proof of extra-sensory perception. It might have reflected his interest in dreams and prophecies and coincidences, but certainly was a case where, for him, open-mindedness had to come before anything else; what was so had to come before what it was convenient to think. On the other hand, he could not make light, as less well-informed people could, of the inconsistency of these ideas with the principles of causality embodied in the existing ‘laws of physics’, and so well attested by experiment.
The idea of ‘teaching’ the machine had also progressed since 1948. By now he had probably learnt by trial and error that the pain and pleasure method was appallingly slow, and had worked out a reason why, which cast a look back to Hazelhurst:
The use of punishments and rewards can at best be a part of the teaching process. Roughly speaking, if the teacher has no other means of communicating to the pupil, the amount of information which can reach him does not exceed the total number of rewards and punishments applied. By the time a child has learnt to repeat ‘Casabianca’ he would probably feel very sore indeed, if the text could only be discovered by a ‘Twenty Questions’ technique, every ‘NO’ taking the form of a blow. It is necessary therefore to have some other ‘unemotional’ channels of communication. If these are available it is possible to teach a machine by punishments and rewards to obey orders given in some language, e.g. a symbolic language. These orders are to be transmitted through the ‘unemotional’ channels. The use of this language will diminish greatly the number of punishments and rewards required.
It was a nice touch of self-reference to bring in Casabianca, for the boy on the burning deck, executing his orders mindlessly, was like the computer. He went on to suggest that a learning machine might achieve a ‘supercritical’ state when, in analogy with the atomic pile, it would produce more ideas than those with which it had been fed. This was essentially a picture of his own development, stated rather more seriously than in 1948, and a claim that even his own originality must somehow have been determined. Perhaps he was thinking of his series for the inverse tangent function, and the law of motion in general relativity, when he first began to put things together in his mind. This, again, was not a new idea. Bernard Shaw had argued it thus in Back to Methuselah, when Pygmalion produced his automaton:
ECRASIA: Cannot he do anything original?
PYGMALION: No. But then, you know, I do not admit that any of us can do anything really original, though Martellus thinks we can.
ACIS: Can he answer a question?
PYGMALION: Oh yes. A question is a stimulus, you know. Ask him one.
Much of what Alan wrote was a justification of Pygmalion’s argument, which Shaw, champion of the Life Force, had derided.
This time he also offered a very carefully phrased prophecy, made deliberately rather than off the cuff to newspaper reporters.
I believe that in about fifty years’ time it will be possible to programme computers, with a storage capacity of about 109, to make them play the imitation game so well that an average interrogator will not have more than 70 per cent chance of making the right identification after five minutes of questioning. The original question, ‘Can machines think?’ I believe to be too meaningless to deserve discussion. Nevertheless I believe that at the end of the century the use of words and general educated opinion will have altered so much that one will be able to speak of machines thinking without expecting to be contradicted.
These conditions (‘average’, ‘five minutes’, ‘70 per cent’) were not very demanding. But it was most important that the ‘imitation game’ would allow questions about anything whatever, not just about mathematics or chess.
It reflected his all-or-nothing intellectual daring, and it came at an appropriate moment. A first generation of pioneers in the new sciences of information and communication, people like von Neumann, Wiener, Shannon, and pre-eminently Alan Turing himself, who had combined broad insights into science and philosophy with the experience of the Second World War, was giving way to a second generation which possessed the administrative and technical skills to build the actual machines. The broad insight, and the short-term skill, had little in common – that was one of Alan’s problems. This paper was something of a swan song for the primal urge, bequeathing the original excitement to the world before it was submerged in mundane technicalities. As such it was a classic work in the British philosophical tradition. It was a gentle reproof to the ponderous essays by Norbert Wiener, as well as to the reactionary, ‘soupy’ trend of English culture in the late 1940s. Bertrand Russell admired it, and his friend Rupert Crawshay-Williams wrote appreciatively to Alan of how much Russell and he had enjoyed reading it.35
From a philosophical point of view, it could be said to fit in with Gilbert Ryle’s The Concept of Mind, which had appeared in 1949, and which put forward the idea of mind not as something added to the brain, but as a kind of description of the world. But Alan’s paper proposed a specific kind of description, namely that of the discrete-state machine. And he was more the scientist than the philosopher. The point of his approach, as he stressed in the paper, was not to talk about it in the abstract, but to try it out and see how much could be achieved. In this he was the Galileo of a new science. Galileo made a practical start upon that abstract model of the world called physics; Alan Turing upon that model provided by the logical machine.
Alan himself would ha
ve liked the comparison: he made reference in the article to Galileo incurring the displeasure of the church and the format of his ‘Objections’ and ‘Refutations’ was one of a trial. A year or so later he gave a talk36 on this subject subtitled ‘A Heretical Theory’. He liked to say things like: ‘One day ladies will take their computers for walks in the park and tell each other “My little computer said such a funny thing this morning!”,’ to destroy any sort of sanctimonious forelock-touching to the ‘higher realms’. Or, when asked how to make a computer say something surprising, he answered ‘Get a bishop to talk to it.’ In 1950 he was hardly likely to be on trial for heresy. But he certainly felt himself up against an irrational, superstitious barrier, and his predisposition was to defy it. He continued:
I believe further that no useful purpose is served by concealing these beliefs. The popular view that scientists proceed inexorably from well-established fact to well-established fact, never being influenced by any unproved conjecture, is quite mistaken. Provided it is made clear which are proved facts and which are conjectures, no harm can result. Conjectures are of great importance since they suggest useful lines of research.
Science, to Alan Turing, was thinking for himself.
Untarnished by all the trials and errors surrounding the actual computer installations, sprang out this ‘conjecture’: the achievement by the millennium of something approaching the artificial intelligence that had long been expressed in the myth of Pygmalion. Also emerging fully-formed was the fruit of his thought since 1935 on the discrete-state machine model, on universality, and the constructive use of the imitation principle to ‘build a brain’.
Nonetheless, beneath the assertive surface of the paper lay probing, needling, teasing questions. For this was not tunnel vision. Unlike so many scientists, Alan Turing was not trapped within the narrow framework within which his ideas were formed. Polanyi was keen on pointing out the different models employed by the different branches of scientific enquiry, and the importance of distinguishing them. But Edward Carpenter had gone to the heart of the matter long before:37
The method of Science is the method of all mundane knowledge; it is that of limitation or actual ignorance. Placed in face of the great uncontained unity of Nature we can only deal with it in thought by selecting certain details and isolating those (either wilfully or unconsciously) from the rest.
To model the activity of the brain as a ‘discrete controlling machine’ was a good example of ‘selecting certain details’, since the brain could, if desired, be described in many other ways. Alan’s thesis was, however, that this was the model relevant to what was called ‘thinking’. As he said a little later,38 in a parody of Jefferson’s argument, ‘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, and so it can’t think”.’ Or as he wrote in this paper,
We do not wish to penalise the machine for its 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 ‘witnesses’ can brag, if they consider it advisable, as much as they please about their charms, strength, or heroism, but the interrogator cannot demand practical demonstrations.
There could be arguments about his thesis within this model, or there could be arguments about the model. The discussion of Gödel’s theorem was, par excellence, one which accepted the model of a logical system. But alive to the philosophy of science, Alan discussed the validity of the model itself. In particular, there was the fact that no physical machine could really be ‘discrete’:
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 switches 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.
That ‘forgetting about them’ would be precisely the element of ‘selecting certain details’ necessary to the scientific method. He conceded that the nervous system was itself continuous, and therefore
certainly not a discrete-state machine. A small error in the information about the size of a nervous impulse impinging on a neuron, may make a large difference to the size of the outgoing impulse. It may be argued that, this being so, one cannot expect to be able to mimic the behaviour of the nervous system with a discrete-state system.
But he argued that whatever kinds of continuous or random elements were involved in the system – as long as the brain worked in some definite way, in fact – it could be simulated as closely as one pleased by a discrete machine. This was reasonable since it was only applying the same method of approximation as worked very well in most applied mathematics and in the replacement of analogue by digital devices.
Natural Wonders had begun by proposing the question, ‘What have I in common with other living things, and how do I differ from them?’ Now Alan was asking what he had in common with a computer, and in what ways he differed. Besides the distinction of ‘continuous’ and ‘discrete’, there was also that of ‘controlling’ and ‘active’ to consider. Here he met the question as to whether his senses, muscular activity and bodily chemistry, were irrelevant to ‘thinking’, or at least, whether they could be absorbed into a purely ‘controlling’ model in which the physical effects did not matter. Discussing this problem, he wrote:
It will not be possible to apply exactly the same teaching process to the machine as to a normal child. It will not, for instance, be provided with legs, so that it could not be asked to go out and fill the coal scuttle. Possibly it might not have eyes. But however well these deficiencies might be overcome by clever engineering, one could not send the creature to school without the other children making excessive fun of it. It must be given some tuition. We need not be too concerned about the legs, eyes, etc. The example of Miss Helen Keller shows that education can take place provided that communication in both directions between teacher and pupil can take place by some means or other.
He was not dogmatic about this line of argument. At the end of the article he wrote (perhaps so as to be on the safe side):
It can also be maintained that it is best to provide the machine with the best sense organs that money can buy, and then teach it to understand and speak English. This process could follow the normal teaching of a child. Things would be pointed out and named, etc. Again I do not know what the right answer is, but I think both approaches should be tried.
But this was not where he placed his own bets. Later he went as far as to say:39
… I certainly hope and believe that no great efforts will be put into making machines with the most distinctively human, but non-intellectual characteristics, such as the shape of the human body. It appears to me to be quite futile to make such attempts and their results would have something like the unpleasant quality of artificial flowers. Attempts to produce a thinking machine seem to me to be in a different category.
In the subjects proposed for automation in 1948, he had been careful to choose those which involved no ‘contact with the outside world’. Chess playing, pre-eminently, would involve no relevant fact but the state of the chessboard and the state of the players’ brains. The same could certainly be claimed of mathematics, and indeed of any purely symbolic system, involving anything technical, any matter of technique. He himself had included cryptanalysis in this scope, but hesitated over language translation. The Mind paper, however, boldly extended the range of ‘intelligent machinery’ to general conversation. As such it was vulnerable to his own criticism, that it would require ‘contact with the outside world’ for this to be possible.
He did not meet the problem that to speak seriously is to act, and not only to issue a string of symbols. Speech may be uttered in order to effect changes in the world, changes inextricably co
nnected with the meaning of the words uttered. The word ‘meaning’ led Polanyi into extra-material, religious connotations, but there is nothing at all supernatural about the mundane fact that human brains are connected with the world by devices other than a teleprinter. A ‘controlling machine’ was to have physical effects ‘as small as we please’, but speech, to be audible or legible, has to have a definite physical effect, tied into the structure of the outside world. The Turing model held that this was an irrelevant fact, to be discarded in the selecting of certain details, but the argument for this irrelevance was left weakly supported.
If, as Alan Turing himself suggested, knowledge and intelligence in human beings derive from interaction with the world, then that knowledge must be stored in human brains in some way that depends upon the nature of that interaction. The structure of the brain must connect the words it stores, with the occasions for using those words, and with the fists and tears, blushes and fright associated with them, or for which they substitute. Could the words be stored for ‘intelligent’ use, within a discrete-state machine model of the brain, unless that model were also equipped with the brain’s sensory and motor and chemical peripheries? Is there intelligence without life? Is there mind without communication? Is there language without living? Is there thought without experience? These were the questions posed by Alan Turing’s argument – questions close to those that worried Wittgenstein. Is language a game, or must it have a connection with real life? For chess thinking, for mathematical thinking, for technical thinking and any kind of purely symbolic problem-solving, there were arguments of great force behind Alan’s view. But in extending it to the domain of all human communication the questions he raised were not properly faced, let alone resolved.
Alan Turing: The Enigma: The Book That Inspired the Film The Imitation Game Page 66