From: Richard Loosemore (rpwl@lightlink.com)
Date: Mon Sep 12 2005 - 10:02:34 MDT
Interestingly enough, a real thinking system (I purport to be one of
those), would respond in the following way to this statement:
"I can easily generate situations in which observing a non-black
non-raven can generate evidence favoring the hypothesis "All ravens are
black" over its alternatives."
... by noting that the question of whether all ravens are black is best
answered, not by fabulously complex appeals to probability theory, but
to some background understanding of what makes them black - genetic
characteristics. The quick conclusion would be "most of them probably
are, but it is conceivable that a genetic variation could generate a
non-black one."
Before you jump on this apparently naive observation, note the
implication I am trying to draw here. Real thinking systems, in
practice, obviously cope with the task of gathering new knowledge by
little strategies and appeals to connected knowledge, of the sort
illustrated here, so from the point of view of actually building an AGI
it might be very productive to study those.
[You might argue that the ravens thing is not really about how people
make such decisions at the conceptual level, but about processes going
on at a much more fundamental level of the AGI. My reply to that is
implied in my previous post on the relevance of complex systems - we
don't yet have a clear reason to believe that Bayesian inference would
be necessary or useful or even applicable at that base (local
mechanisms) level, because nobody has clearly analysed the way that the
conceptual level relates to those lower levels.]
Richard Loosemore
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:52 MDT