🚀 go-pugleaf

greg@Quotron.COM (Greg "Maddog" Knauss) writes:
>dmb@rice-chex.ai.mit.edu (David Baggett) writes:
>
>Rice Chex?  Did you get to name your machine, Dave, or did you inherit it?

Actually, rice-chex is just a general-purpose server.  If you want a
challenge, try finding a breakfast cereal (with approprite hyphenation)
that *doesn't* work when you say "ping <cereal>.ai.mit.edu"  :-)

>Interesting, yes, but practical -- probably not.  It's neat to talk about
>GAs and NNs and say, "When..." and "What if..." but I don't think you're
>going to see them in games (or anywhere else) any time soon.  Even the
>"sensory input" of Pac-Man would overload the machine.

That's very much an assumption, and one based on the current state of
the art.  There are possibly ways of simulating generalized thought
without using the conventional GA and NN methods.  I don't think the
information overload is inherent in the problem, but rather a
side-effect of the solution methods we currently have.

>	What I would like to see (and probabaly should do, with "Night
>of the Brain-Sucking Undead Businessman From Beyond the Grame" if I ever
>get off my butt and start working on it again...  Heh) is fuzzy logic.

That still sort of seems like "faking it" to me.  The character is
still chosing from a predetermined set of choices.  It only becomes
interesting, IMHO, when the algorithm (method) starts producing
patterns of behavior that were not programmed in at the outset.  GA's,
at least, provide a straightforward way to make this happen, though
it's unclear (to me at least) that they have anything at all to do
with independent thought.

Dave Baggett
--
dmb@ai.mit.edu           MIT Artificial Intelligence Laboratory                  
ADVENTIONS: interactive fiction (text adventures) for the 90's!
dmb@ai.mit.edu *** Compu$erve: 76440,2671 *** GEnie: ADVENTIONS