[net.philosophy] A topic for discussion, phil/ai pers

paul@ism780.UUCP (05/17/84)

#R:wxlvax:-27600:ism780:20200001:000:1385
ism780!paul    May 15 21:04:00 1984

***** ism780:net.philosophy / wxlvax!rlw /  9:28 am  May 15, 1984

> It seems that it is IMPOSSIBLE to ever build a computer that can truly
> perceive as a human being does, unless we radically change our ideas
> about how perception is carried out.

> The reason for this is that we humans have very little difficulty
> identifying objects as the same across time, even when all the features of
> that object change (including temporal and spatial ones).  Computers,
> on the other hand, are being built to identify objects by feature-sets.
> But no set of features is ever enough to assure cross-time identification
> of objects.

There are 3 kinds of people in the world: one type believe that humans are
fundamentally different than computers; another type believe that humans are
computers only slightly more complicated than the ones in the big AI labs;
the third type (which I happen to belong to) believe that human brains are
fundamentally computers, but that AI research has barely scratched the
surface of revealing what powerful computers they are.

I would think that once objects can be identified in the first place,
"cross-time identification" by computer (to the same extent that humans can
do it) is *relatively* simple.  It's simply a matter of research being
focused (currently) on specific parts of the general AI problem of "machine
perception".

-- Paul Perkins

rrizzo@bbncca.ARPA (Ron Rizzo) (05/18/84)

There's a fourth type of person: one who doesn't know if humans are
machines and who can't hold any of the 3 beliefs.

The "problem of object identification" is merely a synonym for the
"problem of machine perception".  The adequacy of a scheme of descrip-
tion for an object depends on what you want the machine to DO with that
object.  The problem is we not only lack ingenious descriptive schemes
for objects for performing even simple perceptual
for even simple perceptual skills, but we don't really know (in explicit
rigorous terms) what these skills are that make human perception possible.