From newshub.ccs.yorku.ca!torn!cs.utexas.edu!swrinde!elroy.jpl.nasa.gov!ames!agate!spool.mu.edu!caen!uvaarpa!murdoch!Turing.ORG!lfoard Thu Oct  8 10:10:59 EDT 1992
Article 7092 of comp.ai.philosophy:
Newsgroups: comp.ai.philosophy
Path: newshub.ccs.yorku.ca!torn!cs.utexas.edu!swrinde!elroy.jpl.nasa.gov!ames!agate!spool.mu.edu!caen!uvaarpa!murdoch!Turing.ORG!lfoard
>From: lfoard@Turing.ORG (Lawrence C. Foard)
Subject: Re: AI rights 
Message-ID: <1992Oct2.123153.22335@murdoch.acc.Virginia.EDU>
Sender: usenet@murdoch.acc.Virginia.EDU
Organization: The Turing Project, Charlottesville Virginia.
References: <1992Oct1.232114.1593@murdoch.acc.Virginia.EDU> <1992Oct2.060403.29588@netcom.com>
Date: Fri, 2 Oct 1992 12:31:53 GMT
Lines: 51

In article <1992Oct2.060403.29588@netcom.com> nagle@netcom.com (John Nagle) writes:
>      It's not clear that all sentient systems should have rights.
>It's too early to really address this issue, but it's worth considering
>that the motivational system of an AI is somewhat arbitrary.  Some
>motivational systems are sufficiently incompatible with human society
>that coexistence will not be workable.

Although we do get to build AI's so its best to avoid building psychotic
ones....

>      Systems which are evolved, like Tierra, may present a problem.
>Forced evolution in hostile environments may result in systems with
>very strong survival-at-all-costs imperatives.  These may not be 
>something we want around.

But thats how humans got here, atleast they wouldn't be any worse....
(thats not saying much)

>      Nagle's prediction for the year: before the '90s are out,
>someone will have combined something like Tierra with something
>like the Morris worm, and the result will be a big headache for
>many people.

Real evolving computer worms, well atleast it might make the creationists
think twice.

>      As for the rights issue, for the time being, I suggest a rule of thumb.
>If you believe Moravec (in "Mind Children"), an ant has about 10 MIPS,
>a mouse has about 10,000 MIPS, and a human has about 10,000,000 MIPS.
>(Don't argue about this until you're read the book, please.  He may
>be wrong, but he does have a rational basis for deriving these numbers.)
>So I suggest that we take the position that we can ignore the rights
>issue for anything below a few thousand MIPS.  

This is about the limit of current super computers.
Rough estimate 100,000 machines on internet*20 mips=2,000,000MIPS....
Well all you need is a good neural net program :-)

>But as we advance
>well beyond that level, it may become a real issue.

I don't think its that far away, infact if the numbers you gave are right
its definitly possible now. I can just imagine a neural net morris virus
released over Christmas vacation...

-- 
------          Join the Pythagorean Reform Church!               .
\    /        Repent of your evil irrational numbers             . .
 \  /   and bean eating ways. Accept 10 into your heart!        . . .
  \/   Call the Pythagorean Reform Church BBS at 508-793-9568  . . . .
    


