Why Can't We Write a Grandmaster's Chess Program?

David Deutsch says use only the old theory and you will say AGI is impossible in that way. Know what we know and you will see why it is possible. The people expecting AGI are not making the claim. They say it will just happen with more compute, more computer power.

When chess engines first started defeating the best grandmasters, some people said machines would never do it because humans have a special thing in their brains. Others said it was just a matter of more computer power, and they had a theory of why more power was required. A chess engine and a grandmaster run radically different kinds of computation. The grandmaster does only like one millionth of the computation the chess engine does, and still reaches a very high level, not quite as high as the program. We know how to write the program of a chess engine. We do not know how to write the program of a grandmaster.

A grandmaster can write a book on greatest games and how they were won. Asked how it won, a chess engine traces one move after another, down however many steps it can do nowadays. A grandmaster cannot follow the game down anywhere near as many steps, and will talk about giving up a pawn to create a weakness on the G file. In chess, the side betting on more computer power had a theory of why more power was enough. Deutsch says the claim that AGI will just arrive from more computer power is a different kind of claim.