(ps 920927)
the following was meant to go to the list. i've included it verbatim
below, and followed it w/ my comments.
···
From: cjoslyn@bingsuns.cc.binghamton.edu (Cliff Joslyn)
Subject: Re: turing test
To: sibun@parc.xerox.com
Date: Wed, 23 Sep 1992 12:47:57 -0700
In-Reply-To: <9209230158.AA15492@bingsuns.cc.binghamton.edu >; from "Penni
Sibun" at Sep 22, 92 7:00 pm
(penni sibun 920922.1400)
but what is the criterion for intelligence? the turing test
presupposes that the only criterion for intelligence is being judged
intelligent.
That's fine, but then the TT is not really a test at all. If all it
says is "if someone (you, me, we all) says it's intelligent, then it's
intelligent", then the TT is not even a heuristic. The whole purpose
of a test for X is to distinguish between what merely APPEARS to
satisfy X and what REALLY staisfies X. We don't talk about a test for
redness; under normal circumstances, the perception of redness is its
own test. The TT helps us understand when we're PERCEIVING
intelligence. But like redness, that perception is intuitive, and
automatic; and frequently faulty.
As you've described it, the TT is not a TEST for intelligence, but
rather a DEFINITION of intelligence. That's OK, but let's not get
confused. Of course all definitions are necessary and sufficient;
that's what a definition IS.
(the test can make mistakes,
generating both false positives and false negatives, but that's an
orthogonal issue.)
No, that's not orthogonal; that's exactly my point. The concept of a
false positive or negative assumes that there's some criteria for
Xness independent of the test for Xness. The test provides imperfect
evidence for Xness, perhaps accurate, perhaps not. When the test is
USED, then it provides us all the information we have; but when it's
DESIGNED, then we have the "higher" definitinoal level as well. Is the
TT the criteria, or the test? If it's the test, then what's the
criteria (the prior definition?) If it's the criteria (as I say it
is), then what's the REAL test?
i'd bet
you a lot (if i had it, which i don't!) that you will never ever
discover such a definition that works in the real world. i don't
think you should be spending your time looking for it.
On the contrary, I think the TT IS such a definition, and that it will
work in the real world.
O----------------------------------------------------------------------------->
Cliff Joslyn, Cybernetician at Large, 327 Spring St #2 Portland ME 04102 USA
Systems Science, SUNY Binghamton NASA Goddard Space Flight Center
cjoslyn@bingsuns.cc.binghamton.edu joslyn@kong.gsfc.nasa.gov
V All the world is biscuit shaped. . .
================================================================
> (penni sibun 920922.1400)
>
> but what is the criterion for intelligence? the turing test
> presupposes that the only criterion for intelligence is being judged
> intelligent.
That's fine, but then the TT is not really a test at all.
random house has a concise definition of ``test'':
the means by which the presence, quality, or genunineness of anything
is determined.
i don't see how the tt fails at this.
As you've described it, the TT is not a TEST for intelligence, but
rather a DEFINITION of intelligence. That's OK, but let's not get
confused.
i'm not confused. i don't think you're confused either, though i'm
pretty sure you don't understand what i've been saying.
>(the test can make mistakes,
> generating both false positives and false negatives, but that's an
> orthogonal issue.)
No, that's not orthogonal; that's exactly my point. The concept of a
false positive or negative assumes that there's some criteria for
Xness independent of the test for Xness.
maybe i shouldn't have used the terminology of false positives and
negatives. what i meant was that over time the tester might revise
her judgement, eg, based on new evidence. determining intelligence is
a process of finding out, and you are not going to find out everything
about an entity by the time you have to make a judgement.
you're actually saying tis right here:
The test provides imperfect
evidence for Xness, perhaps accurate, perhaps not. When the test is
USED, then it provides us all the information we have;
but i don't know what you're talking about here. the tt doesn't have
a definition embedded in it.
but when it's
DESIGNED, then we have the "higher" definitinoal level as well.
Is the TT the criteria, or the test?
``criteria'' is plural; ``criterion'' is singular. your
criterion/test distinction seems to add an extra level of abstraction
to how you talk about things w/o having anything more to say about the
phenomena in question (eg, intelligence). you've also set yourself up
for infinite regress, i think.
cheers.
--penni