From: Ben Goertzel (ben@goertzel.org)
Date: Wed Jan 01 2003 - 09:40:42 MST
> My
> point is we don't have AGI until we have a system that can be shown a
> set of learning techniques, which it then applies to new learning
> problems. Also it would have to be able to analogize on those
> techniques, and mutate them, and come up with techniques of its own.
> The Semesters 1 thru 4 may successfully be passed by an non-AGI, but...
> so what? It would be just another Deep Blue. This does not invalidate
> the need for elementary lessons being needed by an AGI. (I know you
> know this, I'm just saying ;>)
>
> Michael Roy Ames
>
Right -- so if people started creating AGI systems with the sole goal of
passing your tests, they'd be missing the point...
A question is: Can you conceive a set of tests with sufficient breadth that,
if an AGI system could pass them, it would be pretty damn clear the system
possessed a high degree of general intelligence. I.e., a set of tests that
can't be viably handled by a narrow-AI system, but yet still possess the
abstract character of your tests, rather than being specialized to the world
of human knowledge or to some particular embodiment...
-- Ben G
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:41 MDT