The BAA goes into some desirable characteristics where
introspection, self-improvement, etc. dominate. We
already have such things in GUIs that adapt to user
habits. Is that self-awareness or environmental
reflex?
It is possible to build such, but I doubt it
qualifies as biological self-awareness. Can
such pass a Turing Test? It depends on the tester,
but for many cases, sure. Now, a lot depends
on who has to be fooled, yes?
It's an interesting BAA for those who like the
challenge of "what they say can't be done"
because even a magnificent failure advances the
state of the art. A fun task is to review the
requirements and try to discern the enervating
simplification (eg, unifying names and locations
in a single string).
len
From: Jonathan Borden [mailto:]
> http://www.darpa.mil/ipto/Solicitations/PIP_02-21.html
>
> Can an artificial system be "self-aware"?
>
Formally define "self-aware" and the answer will be either "yes" or "no"
:-))
