[FOM] The Lucas-Penrose Thesis

[John McCarthy]

This is in response to F.A. Muller's questions.

1. I don't know whether Penrose is still active in arguing against the
possibility of AI, although I don't suppose he has changed his
opinion.

2. In answering objections to his arguments, Penrose doesn't, so far
as I have seen, give references to the criticisms.  This permits him
to express the criticisms as he pleases without any critic being able
to claim that Penrose has misunderstood his argument.

3. I reviewed Penrose's first book The Emperor's New Mind from the
standpoint of an AI researcher in the Bulletin of the American
Mathematical Society in 1989.  My review contains a hypothetical
dialog between Penrose and a computer program, in which, among other
things, the program says it will use any formalization of arithmetic
Penrose likes, provided Penrose will supply confidence in its
consistency.  It also offers to prove Goedel's theorem for any
suitable system.

4. I suppose Penrose read my review, because in his response to
Psycolloquy reviews of his second book he also includes a dialog
between himself and a computer program.  In this dialog he gets the
better of it and crushes the feeble arguments of the program.

5. In the same Psycolloquy issue Sol Feferman finds erroneous
Penrose's arguments appealing to Goedel's theorem.  Penrose gets the
last word and admits technical error but claims the mistake is
patchable.