RE: [agi] A difficulty with AI reflectivity

From: Ben Goertzel (ben@goertzel.org)
Date: Tue Oct 19 2004 - 20:58:52 MDT


Eli:
> Leaving aside other
> qualms I have about the Gödel Machine, it appears to me that for
> the Gödel
> Machine to execute a change that switches to a new theorem-prover or
> proof-verifier, the Gödel Machine must prove that if the proof-verifier
> accepts P, then P.

Hmmm... I haven't read that paper for a while, but my intuitive response is
that you're asking the Godel machine to do more than it needs to do.

Doesn't it just have to prove that:

"If the proof verifier accepts P, then either P or the old proof-verifier
would have accepted P"

or something like that?

Apologies if this answer is off-target, it's a hasty one...

-- Ben G



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:49 MDT