SV: Capacity of the brain

Martin Knopman mknopman at worldnet.att.net
Thu Sep 2 16:26:42 EST 1999


Russell Wallace wrote in message <37CE50A6.66A5 at iol.ie>...
>A couple of other data points: If Moore's Law continues, supercomputers
>will duplicate this processing speed around 2030, with desktop machines
>achieving it by around 2045.  (The storage capacity will be duplicated
>and surpassed well before then.)  It obviously needs to be borne in mind
>that reaching this level of hardware performance is a necessary but not
>sufficient condition for creating a human-level AI.
>
>--

I don't see why you think that reaching this level of hardware performance
is a necessity for human-level AI.  First of all, I don't know what human
level AI means, exactly.  I know quite a few humans who, themselves, would
probably not qualify as human level AI.  Much of the discussion about
intelligence dances around notions of perceived intention which are quite
distinct from anything related to performance.  Just take a look at
different animals and their creative thoughts, which in many cases supercede
almost all humans.  For example, African dung beetles came up with 'the
wheel', essentially (though not the axle).  But they don't seem to qualify
as human-level intelligence, even though some beetle came up with a solution
to a problem that, had a human come up with the same solution he would be
heralded as a great genius.  That the capacity for intention could be
"encoded" is a question quite distinct from hardware or software
performance.

Peace,

Martin Knopman





More information about the Neur-sci mailing list