Newsgroups: comp.ai.alife,comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!rochester!cornell!travelers.mail.cornell.edu!news.kei.com!yeshua.marcam.com!news.mathworks.com!udel!gatech!swrinde!pipex!sunic!news.chalmers.se!news.gu.se!gd-news!d6238.shv.hb.se!sa209
From: sa209@utb.shv.hb.se (Claes Andersson)
Subject: Re: [Q] Robotic 'ants' for Pest control...
Message-ID: <sa209.65@utb.shv.hb.se>
Sender: usenet@gdunix.gd.chalmers.se (USENET News System)
Nntp-Posting-Host: d6238.shv.hb.se
Organization: Department of Scocial Science
References: <39aof0$5ds@todd-06.cs.strath.ac.uk> <39l68m$654@todd-06.cs.strath.ac.uk> <39m9dvINN6de@life.ai.mit.edu> <Cz2641.6xE@watdragon.uwaterloo.ca> <Cz2F6G.AFL@undergrad.math.uwaterloo.ca>
Date: Sun, 13 Nov 1994 23:34:48 GMT
Lines: 43
Xref: glinda.oz.cs.cmu.edu comp.ai.alife:1286 comp.ai.genetic:4255

In article <Cz2F6G.AFL@undergrad.math.uwaterloo.ca> mwtilden@math.uwaterloo.ca (Mark W. Tilden) writes:

>Marvin Minsky <minsky@transit.ai.mit.edu> wrote:
>>
>>As for the COG project of Brooks et al., I disagree with them about
>>the value of building the real-time hardware machinery.  My personal
>>opinion is that more would be learned by making simulated robots
>>operate in relatively simplified simulated worlds.  They object that
>>if you do this, you might overlook serious real world problems.  I
>>don't agree: in my view, it is not important precisely which kinds of
>>noise , you encounter -- or otherwise unpredictable friction effects,
>>etc.  You'll run into the same basic cognotive problems whatever you
>>do, so you might as well introduce cheap, easy to compute types of
>>variation.  I'm not winning that argument, though.

>But isn't there the problem that computer life would then only evolve
>for computer environments?  Isn't the goal to see how to pull
>intelligence out of the box so we can test it personally for validity?
>Granted it might be cool to have a simulated creature on the other side
>of the screen that we could talk to, but so far as been seen there's no
>evidence of intelligence beyond biological means.  Alife environments
>are progressive and neat but they are also too shallow dimensionally to
>emerge competence.

>Computer worlds are fine for us because we can extend our belief to
>encompass them, but without a physical ability, computers have no
>influence to exercise change by themselves.  

>You've gotta build robots, otherwise how can you ever *know*.

>Is all.

 Quite right.. the closest one can come with a computer evolved intelligence 
is one that acts within an emulation of the real world. It will adapt to 
this emulation with all its errors and differences. The problem is that a 
reproducing robot would most certainly take a very long time to evolve to a 
stage where it would be intelligent. One would also have to create a proper 
selection pressure for this to emerge.. But does the primary goal have to be 
intelligence acting like humans when this is impossible? The make 
new instants of life in a "computer-universe" is very interesting in many 
ways.

Claes Andersson. University of Bors. Sweden
