Newsgroups: comp.ai.philosophy,talk.philosophy.misc,talk.religion.newage,alt.atheism,alt.pagan,alt.consciousness,alt.paranormal.channeling,alt.consciousness.mysticism
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!ix.netcom.com!netcom.com!jqb
From: jqb@netcom.com (Jim Balter)
Subject: Re: rereRe: The end of god
Message-ID: <jqbCyo3Lv.4sG@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
References: <36vt2m$g6m@scapa.cs.ualberta.ca> <Cxu1yE.2vL@gpu.utcc.utoronto.ca> <383kau$5q2@scapa.cs.ualberta.ca> <Harmon.672.00119FE3@psyvax.psy.utexas.edu>
Date: Thu, 3 Nov 1994 01:17:55 GMT
Lines: 41

In article <Harmon.672.00119FE3@psyvax.psy.utexas.edu>,
Michael G. Harmon <Harmon@psyvax.psy.utexas.edu> wrote:
>In article <383kau$5q2@scapa.cs.ualberta.ca> kevin@sawnlk.cs.ualberta.ca (Kevin Wiebe) writes:
>
>>.........................................................................
>>"This complete sentence, when given to the machine of description <M> (write
>>it out), will NOT spit it out on the TRUE output slot."
>>.........................................................................
>
>
>>Ok, bend your brain around it for a while, and you will see that it
>>is IMPOSSIBLE for the machine to spit it out EITHER the TRUE or the FALSE
>>slot.  It will choke and die or something.  In any case, without even
>>running the machine or looking at the exact description of the machine, we
>>know the above sentence will not be spit out the TRUE slot (as said above),
>>therefore, the statement itself is TRUE!  We, therefore, can "see" the
>>truth of the statement, but the machine cannot prove it.

[The original posting by Wiebe is gone here, so I'm responding to it as quoted
in Harmon's response.  I apologize for any attribution confusion this causes.]

Perhaps this sort of "thought experiment" is behind the various assertions of
Penrose, Searle, et. al. that humans have capabilities that algorithmic AI's
cannot.  If so, it is a failure of imagination.  Consider what sort of machine
is described by <M>.  It is a machine with TRUE and FALSE slots that is
constrained to spit all true statements and only true statements out of the
TRUE slot, and all false statements and only false statements out of the FALSE
slot.  But why are we interested in such a limited machine, other than to
misleadingly reinforce the claim that we can "see" things that machines (note
the amphiboly!) cannot?  There certainly are <M>s for machines that produce
exactly the same sorts of outputs for such sentences as humans do.  Arguments
for distinctions between humans and axiomatic systems based on *behavior* are
trivially rebutted, since any *specific* human behavior can be emulated.  Thus
the arguments must fall back to mystical claims that humans are made of the
right stuff or that we "know" things that somehow axiomatic systems cannot or
circular "definitions" of consciousness as "... non-axiomatic ...".



-- 
<J Q B>
