Author: GuyHaworth
Date: 07:09:34 10/24/02
Go up one level in this thread
If your focus is man -v- machine, you will be looking for ways of reliably
measuring human -v- silicon capability.
This is bound to involve definitions of 'capability' and the careful creation of
experiments to make the comparison.
Unfortunately, man-machine contests have four inbuilt problems:
1) they are hardly repeatable in a scientific sense
e.g. the 2002 Kasparov is not the 1996 or 1997 Kasparov
2) factors which one might wish to exclude, cannot easily be excluded
e.g. Checkers champion Tinsley suffered from failing health
e.g. environmental parameters which affect man but not machine
3) probably because of '2', large scale experiments are better than small
many man-machine games would be needed to correlate FIDE and SSDF scales
4) in the future, cyborg-concepts might challenge the definition of 'man'
though I guess one could just rule this out
http://www.guardian.co.uk/online/story/0,3605,817484,00.html
is a recent review of the achievements of AI in chess and other domains,
comparing man and machine.
However, without denying the interest of man-machine contests, I think there are
also other goals and challenges:
- to demonstrate that computers can contribute to our lives
... 'added value' rather than 'competitive' challenges
- to create technology of wider use through game-based experimentation
Humans still excel in 'induction', the ability to infer 'a higher meaning' from
the facts around them. Computers seem very poor at this.
The creation of computer-chess programs which can explain, analyse and
commentate with authority in human terms is a greater challenge, and more
valuable and meaningful, than beating Grandmaster K on day D.
g
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.