Re: NEW ARGUMENT AGAINST STRONG AI. A CHALLENGE


My argument does not assume that H's reasoning is sound, it just
assumes that it is alike to ours, alike to the reader's.

Now if Strong AI holds, the reader's reasoning can be reproduced by
some robot, and so can H's. And what a robot (implementing an
algorithmic device) can or cannot do (and not just what it is doing at
same time t) is a physical question, just the same as whether my car
can run 100 miles/hour or it cannot.

So if Strong AI holds, (G') cannot be devoid of truth value.

But then the contradiction seems inescapable.

Put it the following way:

Under some suitable and very common conditions, sentences of the kind
of

(H) the human H cannot believe this sentence

must be paradoxical.

In order to prove it, you only have to suppose that H can perform the
following:

1st. H knows that (H) is true iff H does not believe (H)
2nd. H is aware of his/her own position regarding (H), i. e. if H
believes (H), H knows it, and if H does not believe (H), H knows this
too.

Now, assume (H) is true; then H does not believe (H); but H knows this;
so H has to think (H) is true; so (H) is false. CONTRADICTION.

Assume (H) is false; then H does believe (H); H knows this; then he/she
must think (H) is not true; then (H) is true. CONTRADICTION.

If we remain inside a classical bivalent logic, we must declare (H)
devoid of any truth value, like the Liar etc.

Now suppose H is a Turing machine. The possible relevant behavior of a
Turing machine M is threefold: M stops and writes 1, M stops and writes
0, M does not stop. Then (H) must possess some truth value, for it
refers to a physical/empirical state-of-affairs (the behavior of some
M) that, according to PEM, has either to be or not to be the case.

So, the logical situation is completely different when H is a human (of
the kind required) and when H is a Turing machine, and this shows up in
the fact that we can derive a contradiction if we assume that H is at
the same time a human (of the kind required) and a Turing machine.

This is the core of the argument. Its novelty lies on the fact that it
uses the necessity of a sentence being paradoxical when referred to a
human with very common and usual intellectual capacities.

If there is a reasonable way out of the argument, I really don't know
which.

Regards.

.

Relevant Pages