Quote:
Originally Posted by Piers
Cool.
So a suitable computer program could work out that Goodstein's theorm is true despite the limitations of Peano's axioms, while my window cleaner could'nt
Edit:
OK
Computers are finite, so you are always going to come up with something they cant handle. Gödel common sense or whatever.
Lucas is claiming humans do not have that restrictions. However this is clearly false, just count the neurons, measure someone head whatever.
Penrose makes a similar argument to Lucas. Penrose's argument goes along the lines of:
1) Statements about integers are true or false
2) Humans are sufficiently ingenious to prove such a statement true or false. This may take a while for harder problems such as Fermat's Last Theorem.
3) If you programme a bot to prove theorems, it will be bound by whatever axiom set you typed in initially. So Goodstein's would be unprovable for a Peano bot.
4) Therefore humans are doing something which a bot cannot. Penrose's guess is that this is due to large scale quantum coherence in the brain. This requires a theory of quantum gravity to work.
A lot of people object to many of the steps in this argument. Some people object to the idea that humans doing maths are not algorithmic. Biologists seem to object to the mechanism he suggests.
Scott Aaronsen takes the view that quantum computers cannot solve un-computable problems, but Penrose has human mathematicians doing exactly this.
I guess this is why Penrose says quantum gravity is needed - with current knowledge of quantum theory, quantum computers are potentially quicker but they do not extend the bounds of what is computable. With a theory of quantum gravity, the computable bound could be extended.
I personally think there may something to his argument. It's clear that humans and bots operate fundamentally differently - there are still things where humans are better than bots.