FW: Rape, (2/2)

[rj]

(Found it -- sorry jp. Though the offer to take it offlist, or let it
slide altogether, still stands.)

> 
> I, on the other hand, have very deliberately been all inclusive.

But you don't seem to be including the scenario I have presented, for
one at least. And, as I said, I think it is a scenario which is implicit
in Turing's paper, as also in your comments relating to Turing's paper.

As for Turing's counter-arguments, I'd refer you to No. 3, The
Mathematical Objection:

"In short, then, there might be men cleverer than any given machine, but
then again there might be other
machines cleverer again, and so on."

With the introduction of the adept/non-adept model of human ESP, the
same argument can be turned against Turing's test, viz: There might be
humans (adepts) who can be cleverer than any given machine, but then
again there might be humans (non-adepts) who are not, and so on.

It depends on which is the higher priority for the human adept, shared
humanity, discovering A.I., or concealing his/her adeptness from
non-adepts. Too many variables.

As ESP is not a scientifically-proved fact then you must allow both of
our models to persist. As you must realise, and as Turing certainly
does, it only takes one valid objection to disprove a scientific
hypothesis, and this I believe has been done, and by Turing himself no
less, though he does so, admittedly, grudgingly and apocryphally. This
is why Objection 9 is *not* dismissed: it cannot be dismissed. It is not
the case of waiting for computer technology to get to a certain point;
but of finding out the answer to the ESP question. This is why the
latter question (about whether ESP exists and what form it takes --
which model/s is/are correct) *precedes* the former question about
whether machines can think, or can fool humans into believing they can
think. It doesn't work for me to say that machines can't fool some
humans in some idealised situation, because there is an unknown and
unresolved question in the human side of the algorithm.

"You can fool some of the people some of the time ... " etc

> Back to the Test? Why would either human want to fool the other? They are
> supposed to be in league to debunk the machine. They are bound by the rules
> not to try and fool each other.

Yes. This is the crux. They are "supposed to be" in league but once you
allow the new variable of the existence of ESP then there emerges the
possibility that the two humans will in fact *not* be in league, at
least, not on the issue of the existence of or putative personal
adeptness with ESP. The adept may not wish to disclose his adeptness to
the human non-adept, for whatever reasons; even allowing for the fact
that both humans might still wish to determine whether A.I. exists. This
is why I have said that the ESP hypothesis *precedes* the Turing test
hypothesis, and renders it unworkable. 

> 
> Again, my "model" (and, admittedly, it is evolving as I thrash this out
> with you) would be that "adeptness" is not something that can be possessed-
> but coming into existence only as a shared phenonmenon. 

Yes, I can see this. But as you must realise I have proposed a different
model, equally feasible imo.

> What would be an
> individual trait, would be the willingness to drop the defensiveness and be
> openminded about the possibility of such a mutually uncontrolled link. A
> person with that disposition should be on the lookout for those who might
> want to prey on what they assume to be gullibility. Such predators might
> try to dupe another into believing that a channel existed, when it really
> didn't. So the Turing test can be envisioned as being played one level up-
> between three known humans, two of which are claiming "adeptness"

A non-adept would not have the facility to be "on the lookout", at least
not with anything like 100% accuracy nor 100% of the time, whereas the
adept will never be duped. I'm positing that the channel does exist, but
that it is one way. An adept can dupe a non-adept, but not vice versa. 

In my model the inequality between adept and non-adept might breed
embarrassment, self-consciousness, over-dependence, suspicion, paranoia,
jealousy in the non-adept (just as with any other inequality between
partners). The adept would get sick to death of these after a time as
well. And, again, what of non-volitional thoughts which get through. Not
as simple as saying "Oops, foot in mouth" and retracting a verbal
statement. There's a privacy issue involved, too, I think.

"You would play upon me, you would seem to know my stops, you would
pluck out the heart of my mystery ... " etc 

<snip>
> The duping would
> not be through telepathy, but via the good old fashioned kind of
> human-human interaction.

As long as it's a level playing field. But in this case you are not
allowing for individual differences. What if the ESP adeptness is an
inherited/genetic characteristic, or throwback, like blue eyes; or are
you positing a secret Trystero-style underworld of adepts who have
learnt this skill? I like your model; I think it's great and idealistic.
I'm not convinced that the alternate model I'm positing, which is not so
great or idealistic (but which needs to be considered before we break
out the champagne and crystal balls), isn't equally likely.

Seems we have different models. Yours isn't wrong, but then again
neither is mine. I'm happy with the both/and solution if you are. Until
the scientific fact is patented, that is.

best