I read an article in SciAm (IIRC) about new AI research that picks
from biology - robots with primitive brain-like architectures turn out
to be fast learners, and the hypothesis is that interaction with an
environment is a prerequisite for learning and thus artificial
intelligence.
This might all be old hat to the more academic types here ;-), but my
first thought was that Croquet might form an excellent substitute for
all the hardware work that these guys are currently doing...
(Yup, that's the original question turned inside-out, just wanted to
mention the idea)
On 2/27/06, Alan Grimes <
[hidden email]> wrote:
> I very much look forward to using Croquet as a platform for working on
> an agent based approach to general AI. When a more mature version
> becomes available (grumble grumble), I would like to see what can be
> done with its physics such that meaningful v-physical relationships can
> be explored. I'd also like to explore the possibilities of further
> hardware acceleration such as the ClearSpeed board.
>
>