Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!uunet!mcsun!unido!sinix!es
From: es@sinix.UUCP (Dr. Sanio)
Newsgroups: comp.ai
Subject: Re: Do dogs love their humans (was: Can machines think....)
Message-ID: <1038@athen.sinix.UUCP>
Date: 5 Mar 90 18:18:38 GMT
References: <4030@cbnewsj.ATT.COM> <6791@uhccux.uhcc.hawaii.edu>
Reply-To: es@athen.UUCP (Dr. Sanio)
Organization: Siemens AG, DI ST SP4, Munich
Lines: 49

In article <6791@uhccux.uhcc.hawaii.edu> lee@uhccux.uhcc.hawaii.edu (Greg Lee) writes:
>From article <4030@cbnewsj.ATT.COM>, by jwi@cbnewsj.ATT.COM (Jim Winer @ AT&T, Middletown, NJ):
>>...
>>Language is exactly the point. ...
>
>Now we're cooking with gas.  The Searle-ite intuition is founded on the
>reasoning that `think' is said of humans, computers are not humans, and
>therefore computers can't think.  Or rather, more precisely, computers
>can't `think'.  Apart from the question of word usage, the line of
>reasoning is completely trivial and devoid of any substantive interest.
>
>I stated the fact about word usage too baldly.  It is not exactly true
>that `think' is said only of humans, but rather of things that are human
>or things to which one wishes to attribute some human-like quality.  So
>a more cumbersome but more accurate way of stating the matter is:  if
>you wish to attribute a human-like quality to a computer, you will speak
>of it as `thinking'; but if you are unwilling to make such an
>attribution, you won't.  Opponents of strong AI believe that computers
>are unlike humans, and, in effect, state the consequence of this belief
>for word usage as evidence for holding this belief.
>
So, if it's that easy, why don't you simply redefine "thinking" as "what
computers" (instead of humans) are doing. That way, you would get an 
"intelligent" machine at reasonably low effort. :-)

Back to seriousness: I don't see that the frequently occurring argument
that Searles has no concept of thinking, understanding, intelligence does
lead very far. The problem is that nobody has such a concept.

There are some properties and results of it which can be described, some of
them, like memorizing, recalling from memory, comparing, counting and
computing can be isolated and simulated by several machines, such as books,
meters, calculators, and, most general, computers.

Some time ago, one might have believed to be close to definitely solving the
problem of building a thinking machine when getting one giving reasonable
answers when asked for the cubic root of an arbitrary number.

Nowadays, since that's performed by pocket calculators much faster and more
reliable than every human being could ever do it, the goal seem farther away
then ever estimated. Remember the ability of distinguishing an italic 'a' from
a pica one, a problem which seemed even not to be one, then.

Something heretic: I bet that it will be possible earlier to reproduce a
living (even human) being from the (then) deciphered DNS than building a
machine even passing some kind of Turing test (though I doubt that would be
the breakthru point).
>			Greg, lee@uhccux.uhcc.hawaii.edu
regards, es