From: Eliezer Yudkowsky (sentience@pobox.com)
Date: Sun Dec 29 1996 - 17:10:11 MST
> Wow. I'm impressed. It just goes to show we may not need incredible
> processing speeds to create AI. Perhaps we could analyze the code for this
> program and see if it contains a shortcut to creativity. I know for sure
> that once we get around to perfectly simulating neurons and AI we will then
> be able to create shortcuts, just as numerous vacuum tubes were replaced by
> transisters.
The program in question is called EURISKO. Although you can read about
it practically any popular account of AI, I haven't been able to find
any of the internal details. The creator is named Douglas Lenat, the
same guy who's doing the Cyc project. The basic concept is that the
program operates by using "heuristics" to create ideas, and evaluates
the heuristics to find out which ones worked best. It can also create
new heuristics. This is all done in a fashion similar to the way the
Automated Mathematician (same author, book available) formulated the
Goldbach Conjecture, primes, or numbers, except that Lenat figured out
how to make it work on heuristics instead of just set theory. What
coding he used for heuristics, I have been unable to find.
--
sentience@pobox.com Eliezer S. Yudkowsky
http://tezcat.com/~eliezer/singularity.html
http://tezcat.com/~eliezer/algernon.html
Disclaimer: Unless otherwise specified, I'm not telling you
everything I think I know.
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:35:56 MST