Sean Walsh asks what can a fake Turk chess playing machine teach us about the nature of the mind?
The Turk was a chess playing machine which gained a reputation in the 18th Century for being able to take on all challengers. Rumour has it that its accomplishments include a recorded win against Napoleon himself. No record of that exists, doubtless the Emperor ripped up the scoresheet.
Of course, the “machine” was a fake: prior to every game a human player was smuggled in and was the ultimate source of any “moves” it made.
The story of the Turk can tell us quite a bit about the possibility of “Artificial Intelligence”: that it is possible to get a machine to “think” only by smuggling in (and thereby defeating the point) some version of human agency. And, secondly, that we should not be in default awe of a machine that seems to do things better than we can.
Those who argue that we can get a machine to “think” tend to get things the wrong way around. They point at a particular innovation that looks like it can do what we can (only better) and say: “There you go, nothing mysterious about that. We’re nothing special. Machines are just as good as we are at playing a game of chess or recognising a face in the crowd”. But surely the proper question is: what is the nature of human thought and given that are we certain that what looks like an instantiation is in fact no more than an imitation?
But there is something about the nature of human thought that makes it pretty much inconceivable that “machine intelligence” can be anything other than a pretend version. To believe that a machine can think is equivalent to saying that thought is like a complicated piece of software run on an even more complicated hardware: it is to believe, in other words, that the soul is no more than the mind and that the mind is a program run by the hardware of the brain.
There is a theory of the mind, long popular, that exploits this intuition. It suggests that a “mental event” such as “being in pain” is nothing other than a mediating stage between an input (standing on the plug) and the output (“Ouch!!”). But the account suffers from a certain structural deficiency: what’s important about being in pain is not that it is a mechanism of causal mediation but that it hurts. And that it hurts to me.
The idea that a machine can think is not obnoxious if you are prepared to accept a very uninteresting version of what it means to think. But if we look at the range of things “brought before” the human mind even before we have got ready for work, well, then it seems that the aspirations of the AI proponent are hopelessly unambitious. We wake up, we vaguely remember the dream but not quite, we remember the slight our manager committed against us last week, we worry about the dental appointment, we vaguely recall the last time we slept with a stranger, we then remember that in technicolour and regret the drunken transaction…all that in the first few seconds.
Every moment of our life we interact with other human beings not as machines, not on the assumption that they are other “objects in the world”, but that they are also perspectives on it. There is something irreducibly intimate about a glance in the direction of a beloved that will never be instantiated by a machine because the hyper-manipulation of syntactic structures will never deliver that perspective.
Such is the problem of human freedom: how can we be both objects in the world, governed by the laws of physics, and at the same time subjects in the world – free to act as we like? As Roger Scruton argues in The Soul of the World, our relationship to each other is not one of “object to object” but of “I” to “I” – and the peculiar quality of that relationship is what makes freedom possible. If a machine can think and yet not be aware of itself then it cannot be free and then, as one philosopher once put it, the “artificial” in “artificial intelligence” starts to look very real. Such is the problem of free will, the most perplexing one in philosophy.
And when you run up against problems like that the answer is often wrapped up in allegory. Truth advances behind the cloak of myth. The answer to this problem of how we are both constrained and yet free might be right there in the book of Genesis-the most beautiful combination of scientific and moral allegory ever written. We are both created and free: our freedom is enabled by the fact that we are constrained, rather than is negated by it.
When the Turk chucks the pieces all over the board when he loses then we’ll know he has entered the moral universe of the bad chess player, and therefore the genuine moral universe. And to mine yet further the deeply beguiling world of chess there is a beautiful example of the claim -that freedom is possible within a framework of constraint- that is gifted to us by Samuel Beckett in Murphy. At one point the eponymous anti-hero in that novel plays a game of chess with an inmate in a mental institution, only to find that his opponent is not interested in winning but in subverting the rules of the game. You can find it here. It is worth playing through the moves.
This is what true human freedom consists in: an everlasting reclamation of our freedom carried out within a system of causal constraints. It’s a freedom that no machine will ever have until it knows that it’s a machine.