Author: Ashwin P. Phatak
Date: 09:34:22 10/05/01
Since the early days of AI (Artificial Intelligence) people have agreed on one thing: that a good AI system should mimic or behave like humans, even fail where humans would [Turing test]. However, computer chess programmers did not pay any attention to this - only winning was important. The quest now is only to see who is better - man or machine. I do not doubt that eventually programs like those available today will defeat the best chess players in the world. However, I would rather like to see a program which is "human" - which tries to evaluate positions the way humans do, which makes mistakes, learns from them, and which "enjoys" a hard fought victory. A program which learns how individual opponents play, and devises strategies based on the opponent. A program, which uses psychological tactics. This is what I call Real AI. This is the real goal of AI. The question is not just of philosophical interest - I think if such a program is written, it will help in other areas of AI also. I wonder what contribution the number-crunching approach has made to other fields. It is not general enough and is impossible to apply where the complexity is high.The time is ripe to try a new approach. I am interested in writing a program which uses this "human" approach.If you want to join me,you are welcome. Regards, Ashwin.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.