Who do you think will win the game of Go, AlphaGo or human, in case just before the game start some of the rules of the game are changed/added/deleted?
I don't think alphaGo really has rules programmed into it in the traditional sense. It has a bunch of matrices and a monte carlo search tree. The trainind consists of figuring out which numbers should be in the matrices. If you changed the rules, it would have no capacity to understand such a thing.
Right. Also it would be interesting to analyse how the human player intuition that was developed over period of time playing with old rule will handle the new rules.