I'm double posting this reply to Poetry etc. on somethiing Peter Howard said
about the Turing test for deciding whether a machine has mind. I'm not doing
this because I think my reply is brilliant but only so as not to short change
the Brit po list.
I have always failed to understand why failure to be able to refute is
regarded as evidence either way of the truth of a proposition. Liar's paradox:
I, the liar, am telling you that I'm a liar. True or false?
First, "I am a liar" is a statement about one's "being", whatever that means.
The "am" copula makes this the case. It is not the same as saying, "I always
tell lies."
If the liar is telling the truth then at the moment he/she speaks the sentence
(if his/her whole soul is in the answer -- rational, emotional, spiritual,
etc.) then, for that moment, the liar has changed and is not a liar: the
statement is false. A more normal likelihood is that the whole soul is not in
the answer. In that case the statement is true because he has secretly
remained a liar even while admitting, truthfully, that he is one.
What philosophers will never do is add my parentheses about the "whole soul",
because that would involve other disciplines, such as psychology, and other
beliefs such as metaphysics or even the poetic nature of the imagination.
Therefore, the correct answer is: "The question is inappropriately posed
because we do not have the whole knowledge (state of other person's soul)
which alone can give us the truth and it would be rash of us to make a
judgment meanwhile".
The Turing test proposes that if we can't tell whether or not a machine has a
mind or not because all the evidence it provides is just the same as a mind
would provide, then we have no reason for denying a mind to it. It's a
deviously circular proposition because we define "mind" first by our
experimental reference to a human mind and then, trapped in our own
mindedness, can't think of any test which would make the machine fail. So
therefore the machine is "like us".
But in the Turing test, the possibilities are actually equally balanced: the
machine has been programmed to resemble having mind, or it has mind. If we
know it has been programmed, then even if we have no test to show that it
fails the "mind" qualification, we have discovered no way of deciding between
the two possibilities. If it's a machine, it would be more prudent to reserve
judgment.
OK, we set up artificial circumstances to block the inquirer's knowledge that
he or she is addressing a machine which has been programmed. The inquirer is
convinced that, "The machine has mind". Later, the original inventor is dead,
the machine has been cloaked in android form, everyone who meets it reaches
the same judgment, "It has mind." But to fool the inquirer by creating
ignorance about the true nature of the inquiry is only to define the
conditions in which an ignorant judgment might take place. Most wrong
judgments depend on some kind of ignorance: we do not normally use the
ignorance to justify the correctness of the judgment.
So should we deny that fellow human beings have minds on the grounds that they
may be androids? Oh, well, if you like -- I mean Alice could be. So if I
were really convinced there were convincing androids out there, it would be
prudent to reserve judgment on the grounds of lack of sufficient knowledge.
But if I once knew (sufficiency of knowledge to judge the case) that the
creature I was speaking to was an android, once again I would be unable to
decide whether it had mind or not. If I didn't know there were any androids
about, but there were, well, I should be living in a fools' paradise until
enlightened. But that"s only to say all our beliefs whatsoever are fallible
and we have to come to our judgments on greater or less plausibilities. It
does not make any statement about the nature of mind, except to say that if
someone's claiming to have invented a machine but human mind, that person had
better damn well prove it rather than try to fool me with a whole set of
complicated tests. And as that person would know he/she couldn't do that.
Now, the machine is not testing us. Why not? Well, we could set up a free
programme of some kind where it would learn by its own means and subsequently
*feel* the need to test us. What test could tell us it was feeling? Since
it's not possible to communicate feeling in a testable way, the test would
have to be rationally based. (In their testing of emotions, scientists
usually have to redifine them as "attitudes", which are rational categories
testable by questionnaire.) I don't see, for example, that getting a machine
voice to copy emotional intonations would tell us anything at all about
whether it felt emotion or not, since both kinds of explanation -- it has been
programmed to make that sound/or/ it really feels -- are possible.
The problem with the Turing test is that its judgments are made on rational
bases, yet the mind that is judging cannot be limited to the rationality in
its judgments. Turing, for example, very likely had an emotional bias towards
electing a rational test. You cannot set rationality to discover its own
origins: it is an absolute difficulty. You can get near origin, that's all.
To summarise, the problem with the philosophical/AI question is that it
artificially constricts its experimental environment to exclude those aspects
of mind that it is uncomfortable with -- possibility of intersubjectivity or
telepathy or deep spiritual communion or, perhaps, love that is returned. Oh
well, some lonely person falls in love with a machine which is programmed to
return the love. We are not to use real stupos to work our tests.
Lastly, Peter, the quantum and time jumps. I so easily go wrong if I babble
about science, but as I understand the latest research it is possible to
create experimentally a sort of oscillating system in which two quantum states
of the same particle can be separated, rather as if you could double up its
"being". Obviously, a complicated energy patterning of some kind has been
created. I can't remember the further experiment exactly, perhaps, but it is
something like this: if you change the directional state of one particle
(perhaps in relation to a completely different second particle -- the Herald
Tribune account was confusing) the other one instantly responds without an
*apparent* time interval -- contravening what Einstein was prepared to
believe, but supporting some of the wilder quantum theory speculations.
Let someone better informed clean up my account -- I should check it up in the
journals I suppose.
Doug
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|