Author: Dann Corbit
Date: 10:16:21 04/09/99
Go up one level in this thread
On April 09, 1999 at 12:53:21, Graham Laight wrote: >Excellent post - but I don't quite agree. > >I feel that chess programmers have focused on only a narrow range of AI >techniques. This makes the technology less generally applicable than it could >be. That is true, but it is also a form of information. "What current algorithms can most successfully be applied towards a certain type of searching problem using existing hardware?" is answered heuristically. Since many different methods are tried, we can examine the merits of each. >It has also often resulted in an un-human style of play. Morovic has written a truly excellent book on AI. If we extrapolate computer power increases and if we consider where we are in the current state of affairs, chess programmers are doing it the right way *for now.* To use advanced AI techniques and equal the human brain, we should have the same computational ability of 10^21 operations per second. No computer can approach that right now. >Astonishing accomplishments all the same! Agreed. I also agree that every AI technique should be employed and examined. Not with the goal of winning, but with the goal of seeing the effect. With enough data, we can predict the exact crossover where the more advanced techniques will be superior. Also, since so much energy is spent optimizing the simpler techniques, the more advanced methods have little resources devoted to them. This causes an even more slanted view of their utility.
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.