Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: To Robert Hyatt, Dan Corbit, Christophe Theron , And Other Experts.

Author: Louis Fagliano

Date: 08:54:14 08/05/02

Go up one level in this thread


On August 05, 2002 at 11:10:55, Terry McCracken wrote:

>Do computers make decisions?
>If so, what is your definition of a "computer decision" and how it relates and
>differs from human decisions?
>
>Please cite examples. This can be from chess to any area of so-called "machine
>intelligence", please give _your_ answers, as well as information that can be
>obtained on the net.
>
>Your help with these answers will be greatly appreciated!
>
>Thanks in Advance.
>
>Regards,
> Terry McCracken

You don't need an expert to answer this one.  Computers can't.  The only
"decision" a computer can make is to pick out the largest number in an array of
numbers.  That is why the evaluations of all positions in it's search tree must
be numerical.  (I.e., assign points to certain positive or negative features of
a position such as give yourself 15 points if you have a passed pawn, deduct 10
points if your knight is on the side of the board, etc.)  Then every position in
it's search tree has a numerical value assigned to it and the computer simply
picks the end position (the node) on it's search tree that has the highest
numerical value.  There is no intellegence involved in "picking", for example,
the highest value out of the following array of numbers:

15.2, 4.8, 24.7, -0.9, -2.7, 9.9, 7.9, 10.1

It would be 24.7, of course, and arriving at that "chioce" is not
"intellegence", but a simple mathematical consequence of an assigned function
value.

The intellegence comes from the programmer who must come up with the point
values used in evaluating the position.  This becomes obvious when you consider
the fact that a program can be made to play weaker by changing the parameters of
it's evaluation function (i.e., creating a personality in Chessmaster or Rebel)
If the new parameters make the computer play weaker chess, it doesn't "question"
it's new parameters.  It simply does what it is told.  There is no intellegence.

It may look like intellegence when doing so at high speeds and with a very good
evaluation function (which has to be written by a human), but mimickry is not
the real thing any more than a colored wax figure or mannequin is not the same
as a live person however closely it may resemble it.



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.