Philosophical Notes: The philosopher and the `Chinese room'

ALAN TURING, the Second World War code-breaker, suggested that when we are unable to tell the difference, after prolonged questioning, between whether we are talking to a machine or to a human being, we ought to consider the machine to have intelligence.

This offended many, not just philosophers. After all, intelligence is something hard to acquire and once found, jealously guarded. But it is in the celebrated "Chinese room" experiment that phil-osophers find their champion. There, the artificially intelligent philosopher John Searle sought to debunk such a generous interpretation.

Searle offered to be locked up in an imaginary room with a pile of Chinese symbols. He asks us to consider what would happen if, from time to time, someone outside the room were to post more Chinese ideograms through the letter box for him to sort out, posting back the appropriate symbol, perhaps by referring to some instructions taped on the wall, but written in English.

Searle's argument is that the person in the room does not understand Chinese. This is fairly convincing. After all, at the beginning of his example, he states that they "know no Chinese, either written or spoken", and that for them, "Chinese writing is just so many meaningless squiggles".

This may seem a bit like stating the obvious, but then analytic philosophers do that sort of stuff. The trick is to make the obvious seem not so obvious. The clever bit is that, "from the external point of view - that is, from the point of view of somebody outside the room in which I am locked - my answers to the questions are absolutely indistinguishable from those of native Chinese speakers".

But what seems to have been missed is not that the person in the room appears to understand Chinese, but that the whole "system" - person in the room, sets of symbols on cards, plus instructions taped to the wall, gives the appearance of understanding Chinese. And this is much more plausible. After all, the person who wrote the instructions did understand Chinese. What has happened in his example is that the expertise of the instructions' author has been transferred, via the written rules, to the person in the room.

Broadening the issue, Professor Wang, of Qiangdao University (who really does understand Chinese, unlike Searle) says the question in any case, is not whether the machines demonstrates intelligence, but whether this human construct demonstrates intelligence.

Indeed, a picture, after all, may be said to be "of a tree", or "to demonstrate beauty", or whatever, even if it is basically just bits of mineral on a piece of vegetable.

But this is getting complicated. I should like to suggest instead another "thought experiment" - my own version of this interesting problem. (Searle did several, getting increasingly complicated.)

Suppose a person is locked in a room piled high with dusty old philosophy books. And then suppose on the wall is a blackboard with instructions on how to use them especially to look up views on certain philosophical problems.

Now, into this room are posted some tantalising philosophy questions such as: is evil a normative concept? Are all mathematical truths true a priori? Can machines think in the same sense as people can?

Then using the instructions, our prisoner tears out relevant pages from the philosophy books and posts them back. You see, our prisoner does not understand philosophy. He think it is all just meaningless squiggles. But to anyone outside the room, he appears to understand philosophy.

Alan Turing would say that to distinguish between the appearance and the actualite is mere prejudice - Searle is not so sure. At least, as far as my example goes, we have the option of simply waiting to see if the person gets bored and tries to leave. In which case we can be pretty sure that they don't really understand philosophy.

Martin Cohen is editor of `The Philosopher'