Robots again

From: Adam Jenkins <adam_at_merlin.net.au>
Date: Tue Mar 17 00:30:53 1998

I work in AI from the philosophical side, so I might be able to help here.

>I just caught the end of that special. I didn't realize that interest in AI
>had expired.

Not at all. But there were some problems with teh aims and the possible
realisations. Specifically, the dream of making a computer that could
converse, as per the very foolish Turing Test, has appeared to be further
away than ever. The original dream ran up against the sheer complexity
of the problem, and the way in which the architecture that was being
employed - basically there was a brute strength attitude, which was not
successful. What is needed now is a new direction, which is being
provided in a number of different areas - most notably from Daniel
Dennett's COG.

The one of the big problems was what is know as the Frame Problem of AI.
It's not fitting to explain it here, but in essence the problem is that
it takes time to catalogue the things which are not relevant, as well as
those which are. Thus a computere faced with a problem would have to
spend most of it's time working out what is irrelevant - severly limiting
it's ability to solve anything. There are a lot of other problems, such
as the distinction betweem implicit and explicit memory, but the upshot
is that the old methods failed, and the new are still being developed.

>Was there just no demand for AI, or has the market just taken the useful
>aspects and abandoned the remainder? Isn't the Neural Net technology being
>used in various pattern recognition applications (i.e. OCR) descended from
>the AI research of the the 70's and 80's?

The useful bits are being employed widly - fuzzy logic in toasters,
expert systems all over teh place, OCR stuff, the military make extensive
use of neural nets in the development of new missile technologies, etc.
It has neither been given up, nor proven useless - it's just that we
aren't going to be passing the Turing test for a while yet and, even if
we do, it will prove nothing.

Adam.
Received on Tue Mar 17 1998 - 00:30:53 GMT