From newshub.ccs.yorku.ca!torn!cs.utexas.edu!uunet!olivea!bu.edu!buphy0.bu.edu!leao Mon Aug 24 15:40:54 EDT 1992
Article 6626 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!torn!cs.utexas.edu!uunet!olivea!bu.edu!buphy0.bu.edu!leao
>From: leao@buphy0.bu.edu (Joao Leao)
Newsgroups: comp.ai.philosophy
Subject: Re: Turing Test Myths
Message-ID: <93829@bu.edu>
Date: 17 Aug 92 15:01:03 GMT
References: <BILL.92Aug11105853@ca3.nsma.arizona.edu> <1992Aug12.063425.13479@zip.eecs.umich.edu> <BILL.92Aug12122254@ca3.nsma.arizona.edu> <1992Aug13.024527.2079@news.media.mit.edu>
Sender: news@bu.edu
Organization: Artificial Physics Lab * Boston University
Lines: 126


As long as we are dispelling myths about the Turing Test here are my
2.16 cents:


(1) In spite of the title of Turing's paper it seems to me that it does
not concern itself with whether machines can display intelligence but
(malgre' Minsky) with whether they can _think_ or not. The distinction 
is relevant because 

	(a) Stupidity also requires thought! 
 
 Inconsistent, erratic, poorly structured thought but thought, none-
 theless...
 This is, in rasher terms, what Marvin seems to imply in the first
 part of his paper and is the gist of his counter-argument to 
 the Penrose-Lucas argument. "Inconsistent thought" maybe the royal
 road to mathematical truth, though that does not make it non algo-
 rithmic necessarily...

>minsky@media.mit.edu (Marvin Minsky)

>
>in view of the many mistakes we all
>make, I see no compelling evidence that anyone has any direct such
>access to truth.  All we can depend upon (including the power of
>formal proof) is based on our experience.
>

 That is to say: not all thought is intelligent thought! "Unadmirable
 mental performances" abound (regardless of our incapacity to define
 "though" or intelligence). 


 	(b) It is an intuititively weaker claim that, say, 'animals
	think!` (which some of us may be willing to accept with less
	revulsions) than that 'animals are intelligent!`. 

It is quite possible that Turing was aware of the whole controversy
about animal thought going back to Descartes (as Gunderson and others
have defended) and though that is not explicit in the 1950 article
it is clearly relevant to the argument. Today we have quite a good
deal of evidence of both animal thought and animal intelligence 
(symbolic processing if you will) and none of the people who 
have gathered it would put an animal (gorilla or dolphin) through 
something like an Imitation Game to justify their claims! 

(2) It is clear that Turing's argument is filled with irony but
it is important to aknowledge that he uses that irony to build
what he presumably believed (and a lot of other people still
believe) amounts to a GOOD TEST OF THINKING (I have Dan Dennett
on tape saying those very words)! Because we only have direct
access to our own thoughts we must identify thought in other
entities on the basis of a form of "empathy" well exemplified 
in conversation. When none of the features where we normally
"anchor" that empathy (the fact that most of the entities that
talk to us look like us) than the problem of detecting thought
becomes considerably more intrincate. It is very possible that
we can only do it by trial and error, questioning and comparing
answers until our experience with conversations effectively
reveals something wrong. 

(Personally I do not believe that the Turing test is a good test
of thinking and I have an argument to that effect ...but that is
another story!)

(3) The paradigm of the Imitation Game with Male/Female only
makes sense in Turing's argument if one intuitively realizes 
that it would be DIFFICULT to be the judge on such a game
between humans! Contrary to 

>bill@nsma.arizona.edu (Bill Skaggs)

>I believe that, given an hour of interrogation, I would have a better
>than 90% probability of distinguishing between a man and a woman.  I
>might be right, or I might be wrong, but it doesn't seem reasonable
>that the question whether I'm right or wrong has anything to do with
>the question whether machines can think.

I sincery doubt that you would be able to do this with any consistency,
so would Turing (otherwise he would have phrased his argument differently).
His point is that you do not have any more of a chance to develop a
strategy that will systematically tell between a human and a machine
pretending to be a human on the other side of a teletype than you
have of developing a strategy that would systematically distinguish
a woman and a man pretending to be a woman in the same circunstances!
OK? The common link between the two situations is your incapacity to
draw the line and this is where the strength of the argument lies:
The only standard of machine intelligence is human "stupidity" (or
"uncommon sense" as I would prefer to call the propensity for
inconsistency which Marvin stresses in his article!

>Well, in spite of all the interest the Turing Test has drawn over the
>years, nobody has ever (to my knowledge) actually carried out the
>man-woman imitation test, even though it would be pretty easy to do.
>On the other hand, a man-machine imitation test *has* been carried
>out.  Do you want to say that it's impossible to interpret the results
>of the man-machine test in the absence of data from a man-woman test?
>I just don't believe that this is what Turing meant.

It would surely be interesting to carry out the Imitation Game in its
M/F version but I suspect, as you do, that the outcome in any case 
will only be remotely relevant for the question of machine thought.
After even if you happened to develop a "judgeing strategy" for telling
a woman from a man-pretender (or vice-versa) it is very unlikely that
the same would work on the Machine/Human case...
 
Incidentally, the Loebner Prize competition -- which was readily 
dismissed by the purists in this group -- showed already some of
the features anticipated by Turing, specially in its surprising
results (considering the low expectations of the organizers)! 
By the way I have the transcripts of the dialogues and
would venture to make them ftp-able if people are interested!
Speak up...






-- 
Joao Pedro Leao (Artificial Iconoclast and Director of Computer Resources
Artificial Physics Lab * Boston University - Physics Dept. Boston MA 02215)
 	leao@buphy.bu.edu | leao@buphyc.bitnet | BUPHYC::LEAO 
"Well I am sitting here in Tahiti/ I am laying in the sun and sipping a...
...chartreuse tropical drink!/ and I say: I know those Bermuda shorts!..."


