Bennett and Hacker: Village Idiots or Philosophers?

David Longley David at longley.demon.co.uk
Mon Feb 16 11:26:27 EST 2004


In article <4030e04a.30726781 at netnews.att.net>, Lester Zick 
<lesterDELzick at worldnet.att.net> writes
>
>Hi Eray -
>
>I certainly agree with what you note here. The problem with arguments,
>rationales, etc. is that they are only about as useful as people's
>comprehension of them. I think they are conclusive once understood but
>Neil considers them totally or mostly word salad and you seem to be
>somewhere in the middle.
>
>But I'll say one thing for the arguments, they're brief. So they admit
>of evaluation in pretty straightforward terms. The only complicated
>rationale is for S "differences between differences" resolution of
>Russell's paradox and I'll be posting more on that in a few days.
>
>The unfortunate thing is they don't have any obvious direct relevance
>to immediate issues in ai as the subject stands. The only significance
>I can think of at the moment is that these ideas indicate that the
>idea of actual sentience in ai is really something more than programs
>and whatever one chooses to project as ai in turing terms.
>
>This latter is more on the order of robotics or in cognitive arenas
>what I refer to as artificial neural turologies - ants. Which I find
>nothing wrong with because it will probably prove more useful than
>actual models of general cognition. However as Jim Bromer points out
>in his Re: Reasoning and AI yesterday, it has been the case that
>designers and programmers have thought they were more or less
>discovering and writing equations of cognitive behavior and sentience
>with their programs and that has definitely not proven to be the case.
>So I consider that it would behoove ai architects to understand why so
>they can reconsider whether they are aiming at actual cognition or
>just robotics and the difference between the two.

Go and find out about *discrimination learning*.

-- 
David Longley



More information about the Neur-sci mailing list