From: Ben Goertzel (firstname.lastname@example.org)
Date: Mon Oct 17 2005 - 13:56:40 MDT
> Wasting time on robot engineering is not high on our agenda.
> Brian Atkins
While I agree that humanlike embodiment is probably not on the "shortest
path" to human-level AI, I'm not so certain of this that I would refer to
robot engineering as "wasting time" ;-)
It's obvious that human-level or superhuman intelligence don't intrinsically
require any kind of embodiment, let alone humanlike embodiment.
However, it's also clear that embodiment provides a very handy way for a
mind to learn a lot of things (including how to model itself as part of the
world, how to empathize with others, how to carry out simple plans), as well
as how to communicate with people (e.g. learning the grounding of "over" and
"by" has got to be a lot easier in an embodied context).
In the Novamente project we are trying to take a middle ground by embedding
Novamente in a 3D simulation world rather than hooking it up to a real
robot. And to avoid having to mess too much with vision processing, we are
making Novamente "see" polygons rather than pixels.
This exemplifies what Michael Wilson was saying about the diversity of
attitudes toward embodiment. My own attitude is that embodiment is VERY
USEFUL but not necessary for AGI. Embodiment makes it easier for AI's to
learn some things, but also involves a certain amount of hassle and
distraction from core cognitive issues.
-- Ben G
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:52 MDT