Author: Robert Henry Durrett
Date: 19:05:11 05/15/98
Go up one level in this thread
On May 10, 1998 at 18:51:35, Vincent Diepeveen wrote: >Most definitions of AI require several things which are all met by >a chessprogram. The general exception is the requirement >that a program must learn, this is the problem >for chessprograms, they don't learn well. > >Like if we play another game it makes the same strategical mistakes >it made previous game. It doesn't learn. There doesn't seem to be any reason why a chess machine could not be designed to perform post-mortem analyses of it's games and then make some sort of adjustments [such as parameter adjustments] before the next game. Possibly, no chess machines have been designed to do that, but if not, it may only be that the chess machine designers have not chosen to do that. The way humans learn from their games is to do post-mortem analyses of their games, especially the lost games, and surely this could be emulated by chess engines. [Assuming noone pulls the plug between games.] Whether or not that would merit the title "Artificial Intelligence" is up to the theorists. As a practical matter, ability to learn may merely be a matter of designing the machines to do post-mortem analyses and to adapt based on findings. The types of allowable adaptations would be chosen in advance by the machine designers. Perhaps that's what happened to humans, too.
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.