Re: computer-go: Pattern matching

[Heikki Levanto <heikki@xxxxxx>]

Tristan Cazenave <cazenave@xxxxxxxxxxxxxxxxx> wrote:

> Is AI a boring academic field with people publishing
> a lot of papers about useless theories, or a domain
> where people try to enable programs to behave as
> 'intelligent' as they can ?


An old joke is that AI is making computers do things that require
more intelligence than computers have. Chess used to be near AI,
because it "obviously" required more "intelligence" than was to be
expected from mere computers. Once it had been solved (with mostly
brute force, better hardware, and some clever programming) it ceased
to be counted as AI.

Now, today Go obviously requres far too much computing power to be
practical to solve with brute force. Therefore it is considered a
good AI problem. I believe we will some day see a really strong Go 
program, but at that time consider it "only" to be a clever application
of (that day's) routine programming, and lots of brute force. It may
well be that this program learns a lot, uses neural nets, logic
programming, deduction, intuition, and what have we, but by then
they will be the routine methods for that sort of problems...

Still, today Go is a great challenge, and I enjoy reading about the
various ways people have tried to approach it. I have some ideas of
my own, and as soon as I get the time (heh) I will try some of them
out in practice. If I would happen to hit upon a good solution
(unlikely!) I might tell myself that I have improved the state of 
AI as it is known today (and be mighty proud of that...)

- Heikki

P.S. Is this too philosophical for the compgo list???


-- 
Heikki Levanto     LSD Levanto Software Development   heikki@xxxxxxxxxxxxxxxxx
               "In Murphy we Turst"