Newsgroups: comp.ai.alife,comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!swrinde!emory!mind.org!runic!thantos
From: thantos@runic.mind.org (Alexander Williams)
Subject: Re: [Q] Robotic 'ants' for Pest control...
Organization: Runic Writings UUCP Site - Lawrenceville, GA
Message-ID: <CzD3wE.5Ez@runic.mind.org>
References: <39aof0$5ds@todd-06.cs.strath.ac.uk> <Cz2F6G.AFL@undergrad.math.uwaterloo.ca> <Cz5pE2.4Ez@runic.mind.org> <1994Nov15.155101.20116@dcs.warwick.ac.uk>
Date: Wed, 16 Nov 1994 13:24:12 GMT
Lines: 58
Xref: glinda.oz.cs.cmu.edu comp.ai.alife:1323 comp.ai.genetic:4318

In article <1994Nov15.155101.20116@dcs.warwick.ac.uk>,
Benjohn Barnes <B.Barnes@dcs.warwick.ac.uk> wrote:
>I agree with both of you (I think) but do you need a compex environment for a
>really sentient entity (great word) to develop in? Maybe such an environment
>would be dificult to simulate on a computer.

I suppose  that depends   on  what kind  of  sapience (as  opposed  to
sentience, the ability to  feel, I put  sapience, the ability to think
or  consider) you're  looking for.    If you're  trying to develop   a
battlefield robot who    maneuvers  over multiple  kinds  of  terrain,
acquires and   destroys   targets,   all  in  multiple     environment
conditions... then no, you  can't do /all/ that to  high accuracy in a
computer.  You  can come /close/, by  differing measures of close, but
you can't do it as well as  actually building a  robot and sticking it
out there.  On  the other hand, such a  construct is  very unlikely to
evolve to sapience, actually  considering what its doing with internal
modeling, etc,  (partially because its a Bad  Thing(tm) if it does, it
might decide  your enemy is not its  enemy) while whether its sentient
or not,   feeling,    evaluating,    reacting  based   on    sensation
intelligently, may not be in any doubt.

Now,  let's look at   another gedanken beast.    This one exists  soly
inside, oh, let's be generous and say a  CM5, with the works hooked up
to  it.  And when I  say the works, I mean  the works: digital cameras
around the  lab, microphones,   several newsfeeds  coming in  off  the
satelite, the  entire alt.* hierarchy... you  get the  idea.  Now, for
the sake of  argument, let's  say we  set  the  bunch  of it  on  some
multi-faceted task, produced by multiple GA's,  Agents, NNs, etc, some
of which   analyze incoming  news stories,   create  logical precis of
postings to alt.sex.fetish.foot, respond to voice and gesture commands
from workers in the lab, etc.  The primary stimulation for this Grande
Automaton is intellectual,  datum-oriented.  The only sensorium it has
that you   might label "raw"  is  camera  and  microphone.  Can /this/
Agent, taken in  toto, evolve?  Logically,  yes it can.  What will  it
evolve towards?   Better execution of its  duty,  which, would seem to
encourage its motion toward sapience, toward cognition, internal-state
modeling, predictive  facilities  based on   concept,  etc.  So,  this
entity  would be  /more/  likely to  evolve  to  sapience than  toward
sentience,  in  a  sense,    if   you'll excuse my    science  fiction
proposition-style.   Not   to  say it     might  not develop   a  fine
appreciation for hardcore thrash  metal music, which is pretty clearly
a sapience concern, but it has access to cognitive input as well.

In short, or  at  least, in  conclusion, "complex environment"  within
systems can be as informationally-complex  as  you're willing to  get.
Satelite images, newsfeeds, a world of information humming right here.
If you refer to a physical model of the environment, then yes, limited
complexity is all we have right now,  but even that might be sufficent
to  develop a sentient  Agent, in one  form or another.  What it comes
down to, in the end,  is philosophy, not  science.  When will you call
it sentient?   What does it  take to  be called  sapient?  Do you care
about the difference?

-- 
thantos@runic.mind.org (Alexander Williams)     | PGP 2.6 key avail
                                                | DF 22 16 CE CA 7F
  Do What Thou Wilt Shall Be the Whole of the   | 98 47 13 EE 8E EC
  Law. Love is the Law, Love Under Will. -oOo-  | 9C 2D 9B 9B
