Re: chess.com/news/view/google-s-alphazero-destroys-stockfish-in-100-game-match

AlphaZero Crushes the Game of Chess. Remember how AlphaGo — Google’s Go-playing AI — went 5-0 against the reigning, 3-time European Go champion? Well, Google’s DeepMind team is back with AlphaZero, and it’s just crushed the game of chess. Mike Klein, writing for the Chess.com company blog:

A little more than a year after AlphaGo sensationally won against the top Go player, the artificial-intelligence program AlphaZero has obliterated the highest-rated chess engine.

Stockfish, which for most top players is their go-to preparation tool, and which won the 2016 TCEC Championship and the 2017 Chess.com Computer Chess Championship, didn’t stand a chance. AlphaZero won the closed-door, 100-game match with 28 wins, 72 draws, and zero losses.

Oh, and it took AlphaZero only four hours to “learn” chess. Sorry humans, you had a good run.

Take a minute to notice what’s interesting about this narrative. It’s not that a computer defeated a human, like with AlphaGo, it’s that one computer defeated another computer. At what point will it be more entertaining to watch two AIs compete instead of two humans? While you’re chewing on that future scenario, let me point out why AlphaZero’s accomplishment is being seen as more impressive than AlphaGo’s, despite Go being significantly more complex than chess.

With AlphaGo, Google’s researches, “trained the policy network on 30 million moves from games played by human experts”; essentially, “watch how the best do it, and use that information to choose your moves”. However, AlphaZero wasn’t trained on some huge dataset of chess matches. In fact, AlphaZero wasn’t “trained” at all. Back to Mr. Klein:

This would be akin to a robot being given access to thousands of metal bits and parts, but no knowledge of a combustion engine, then it experiments numerous times with every combination possible until it builds a Ferrari.

At least we still have physical sports. For now.

Sunday, 17 December 2017