[FOM] AI in Chess/Kasparov

Harvey Friedman friedman at math.ohio-state.edu
Sat Jan 30 19:46:21 EST 2010


The above touches on some interesting AI issues.

Most interesting is just how much better human + computer is than  
human alone or computer alone.

This is quite useful - it prima facie refutes the worry that humans  
playing humans are just fooling each other into thinking that they are  
any good.

(I am not sure that correspondence play time limits were taken into  
consideration in the article, however. That might add a new dimension  
to the discussion).

Also to ponder:

amateur human + skilled computer use dominates top human + competent  
computer use.

This all seems to suggest that computers and humans have orthogonal  
abilities. I wonder what, if anything, rigorous can be learned from  
the chess experience.

Consider the following Conjecture in three forms:

A Human chess player will come along whose ideas are so creatively  
powerful, that

i. (s)he will consistently beat humans.
ii. (s)he will consistently beat computers.
iii. (s)he will consistently beat human + computer.

I wonder what people think of these Conjectures, and how we may try to  
say something interesting about them rigorously, before the GREAT ONE  

Harvey Friedman

More information about the FOM mailing list