From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!sun-barr!decwrl!mcnc!aurs01!throop Mon May 25 14:05:29 EDT 1992
Article 5662 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.ecf!utgpu!cs.utexas.edu!sun-barr!decwrl!mcnc!aurs01!throop
>From: throop@aurs01.UUCP (Wayne Throop)
Newsgroups: comp.ai.philosophy
Subject: kidneys and fish
Summary: no, not a new flavor of cat food...
Message-ID: <60689@aurs01.UUCP>
Date: 14 May 92 18:56:32 GMT
References: <1992May10.041234.8885@ccu.umanitoba.ca> <1992May11.163332.27781@psych.toronto.edu> <1992May13.001033.14320@ccu.umanitoba.ca> <1992May13.182043.40913@spss.com>
Sender: news@aurs01.UUCP
Lines: 33

> markrose@spss.com (Mark Rosenfelder)
>> zirdum@ccu.umanitoba.ca (Antun Zirdum)
>>I suppose also that functionalism demands that a kidney be
>>implimentable in a school of fish (actually I'm sure of it!),
> Boy, you just can't tell the players without a scorecard.  Do you realize
> you're supporting Searle, here?  You're saying that the difference between
> a physical kidney and a simulation of a kidney is the same as that between
> "real intelligence" and intelligence simulated on a computer (in this case,
> a school of fish).  That's Searle exactly; see Sci. Am. 1/90, p. 29.

I think "functionalism demands" that we realize that that school of
fish is just not a good enough simulation.  A dialysis machine, now
THERE's a simulation of a kidney that's good enough (in some ways).

Now, that dialysis machine isn't "being a kidney" be virtue of
simulating one, but then nobody is claiming that an AI would
"be a human" by virtue of simulating human thought.  The dialysis
machine is simulating a kidney's filtering function, and the
AI is (supposedly) simulating a human's mental function.

Just as the dialysis machine really does filter blood by simulating
the kidney's method, I see no convincing reason to conclude that an AI 
might not really think by simulating the human method.

> However, if an implementation of an algorithm is all that's necessary for
> intelligence, *any* implementation will do, and has "real" intelligence.

I agree, but have reservations about what might be meant by an
"implementation of an algorithm".  Putnam's-Rock-like notions of
implementation of an algorithm, for example, don't seem to really
capture the notion.

Wayne Throop       ...!mcnc!aurgate!throop


