3 ms·
What I find fascinating - and I guess this really highlights that I have no idea whatsoever how AlphaGo works - is that at the start of game 2, AlphaGo plays P4
by oneeyedpigeon 11y ago
What I find fascinating - and I guess this really highlights that I have no idea whatsoever how AlphaGo works - is that at the start of game 2, AlphaGo plays P4, then Lee Sedol plays D16. To a layman, this looks like it would be a very, very common opening. Moreover, it's symmetrical - I'm not sure how that affects things, but my naive intuition is that it makes the game state less complex.
Nonetheless, AlphaGo takes a minute and a half to play its next move. Can anyone explain what on earth is going on during those 90 seconds?
- edraferi 11y agoMy guess is it's doing the same thing it does every move: look at the state of the board and calculate the next move. A person might memorize standard openings and play them by rote. The computer _could_ do that, but it would just be a computational short cut. In a sense, the machine is re-discovering the classic opening.
- versteegen 11y ago90 seconds per move was normal for AlphaGo, wasn't it? I didn't pay attention to how its time per move varies, but it's a question I would love to ask the team (although, maybe it's explained in the paper?) You have to realise that it's not using an opening book, it's computing the best response from scratch.