Ground Zero: DeepMind Drops an Alpha-Bomb on the Chess Community
Six months ago, as competitors in the London Chess Classic prepared for their matches, rumours began circling amongst the international chess community that the best chess program in the world, Stockfish, had just been beaten. By an A.I. system that had taught itself chess in only four hours with no access to opening books or endgame tables.
So much for humans spending four decades trying to become a GrandMaster!!!
Developed by research company DeepMind, AlphaZero is a computer program that utilises deep neural networks (information processing paradigms inspired by the way biological neural systems process data) and custom tensor processing units (A.I. accelerator application-specific integrated circuits developed by Google specifically for neural network machine learning) to self learn.
Trained solely through self-play reinforcement learning, starting from random play, and without any supervision or use of human data, after a mere four hours of self-training, AlphaZero was estimated to be playing at a higher ELO rating than Stockfish. After nine hours, it decisively defeated Stockfish in a time controlled 100 game tournament (28 wins, 0 losses and 72 draws).
“It doesn’t play like a human, and it doesn’t play like a program,” said Demis Hassabis, the founder and CEO of DeepMind.“It plays in a third, almost alien, way.”
Hassabis speculates that because AlphaZero teaches itself, it benefits from not following the usual approach of assigning value to pieces and trying to minimise losses. “Maybe our conception of chess has been too limited,” he said. “It could be an important moment for chess. We can graft it into our own play.”
What’s really interesting is that AlphaZero sometimes makes seemingly crazy sacrifices, like offering up a bishop and queen to exploit a positional advantage, emerging victorious. This goes against the grain as usually in chess, such sacrifices of high-value pieces are rare.
The following video shows one of the 100 games AlphaZero played against Stockfish and demonstrates the A.I. system’s unique approach to chess.
Top chess GrandMasters were blown away by AlphaZero’s capabilities.
“Of course the result is extremely impressive; I wouldn’t even dream of winning one game against Stockfish.” Maxime Vachier-Lagrave
“I was amazed. I don’t think any other engine has shown dominance like that. I think it was four hours of learning so who knows what it can do with even more.” Fabiano Caruana
“I am very much surprised because we normally work with Stockfish and it looks like it’s a good program but if we have a program which beats Stockfish so easily it might be a new generation for computers and maybe it’s a historical day for chess. We’ll see how it will get stronger!” Sergey Karjakin
“I was shocked. This is the new big thing. It totally changes chess. It might be rated, what, 3700? Close to 4000? That’s really crazy.” Wesley So
“I guess we’ll just start preparing with it. It’s hard to know the ramifications of opening developments. I guess there will be some opening developments and maybe we’ll find out that some opening lines aren’t as good as we previously thought … It will probably advance our knowledge of chess a bit and it will probably take many years before we get to that next level.” Caruana
Though one Master wasn’t too concerned about how it would alter his game …
“With my lazy preparation it probably wouldn’t help too much; I’ll never be able to remember anyway!” Adams
What do you think about AlphaZero and its potential to change chess theory, strategy and gameplay?