AlphaZero AI Beats Champion Chess Program after Teaching Itself in Four Hours

Friday, 08 Dec, 2017

DeepMind, Google's artificial intelligence programme, has not only surpassed the limits of human ability in chess, but it has done so in only four hours, according to The Week.

Details of the most recent tests were published on Wednesday (6 December) by New York's Cornell University, but findings are still to be peer-reviewed. However, AlphaGo researchers experimented with the program's ability to teach itself to play, without any need for human feedback, which ultimately translated to chess, too.

Going up against AI system Elmo at shogi, AlphaZero won 90 games, drew two and lost eight.

Developed by Google's DeepMind AI lab, AlphaZero is a tweaked, more generic version of AlphaGo Zero, which specialises in playing the Chinese board game, Go.

DeepMind achieved the feat learning from scratch with its AlphaZero programme which has "superhuman" knowledge. The new AI's programming did not include data for tactics or strategy specific to these games; rather, it was provided with some general rules and instructions for the game, no different than what a human might be provided their first time attempting chess or Go.

"It will no doubt revolutionise the game, but think about how this could be applied outside chess".

Some publications are reporting that AlphaZero "taught itself how to play [chess] in under four hours", but that's not entirely accurate. When the exercise starts, AlphaZero already knows how to play chess - just not very well. The system can churn out 800,000 positions each second, as compared to Stockfish 8's 70 million moves a second.

AlphaZero was also pitted against its sibling, AlphaGo, which was also modified to play chess.

"I always wondered how it would be if a superior species landed on Earth and showed us how they played chess", grandmaster Peter Heine Nielsen told the BBC. Only time will tell.

Chess enthusiasts have been impressed with the program's immediate dominance. In the 1990s, Russian chess master Garry Kasparov was defeated by IBM's supercomputer, Deep Blue. It may be studying fewer positions, but it's looking for the most promising ones, an approach that worked against Stockfish. But by adapting the system to learn a new set of rules for an entirely new game, the DeepMind developers demonstrated the flexibility of the system and (possibly) its potential to work outside of mere gameplay. The AI system's latest success is within this vein, but in reality is about so much more than that.

AlphaZero defeated Stockfish in 25 games, in which it had white pieces.