Programming Go

from the not-so-easy dept

Back when Deep Blue became a chess champion people were talking about how computer AI was finally catching up with the human brain. However, some felt that wasn't true, since Deep Blue was merely running through many different options extraordinarily quickly - and not doing any kind of "thinking". The real challenge, according to some, is to create a computer program that plays Go (NY Times registration - generate a fake one). I've been a terrible Go player on and off for many years, and recently started playing around with a few shareware Go products, and found them mostly lacking. This article points out that the difficulties in creating good Go playing software go well beyond the number of calculations it has to perform and towards real pattern matching - which is what AI is supposed to be about. It sounds like it may be quite some time before we have a champion Go computer player.