RE: AI and Moore's Law redux

From: Ben Goertzel (ben@goertzel.org)
Date: Thu Jan 31 2002 - 23:31:19 MST


> Either way, I am definitely awaiting Wolfram's new book (expected in
> March) to see what he has to say about it! (Have you looked at
> wolframscience.com yet?) Some of the publisher's claims about the book
> sound rather grandiose to me, and the only reason I tolerate these is
> because, heck, he's Stephen Wolfram. :>

It is pretty clear to me what Wolfram has done here.... I can't be
sure how well he's done it until I read the book of course.

He's shown that *qualitatively*, a hell of a lot of complex-systems
phenomena -- including ones that are generally reminiscent of particle
physics, the origin of life, human memory, etc. etc. -- can be gotten
to emerge from simple 1D cellular automata.

He has not made detailed accurate models of real-world phenomena using
these 1D CA's. Nor has he created a general abstract rigorous math
theory of what kinds of 1D CA's will generate what kinds of complex
phenomena. Nor has he created an AI or anything of practical value
using these 1D CA's.

Rather he has assembled an amazing panoply of fascinating examples.

I tend to agree with his intuition that this kind of thing -- complex
patterns emerging from simple computational systems -- is going to be
the science of the future. But I suspect that we need to prove strong
theorems about these systems, not just compute fascinating examples; and
I suspect it will be AI-driven automated theorem-proving systems that
prove these theorems...

-- Ben G



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:37 MDT