Popular Mechanics (South Africa)
AI’s undisputed chess champion
IT WAS A war of titans you likely never heard about. A year ago, two of the world’s strongest and most radically different chess engines fought a pitched, 100-game battle to decide the future of computer chess. On one side was Stockfish 8. This world-champion program approaches chess like dynamite handles a boulder – with sheer force, churning through 60 million potential moves per second. Of these millions of moves, Stockfish picks what it sees as the very best one – with ‘best’ defined by quite a complex, hand-tuned algorithm co-designed by computer scientists as well as chess grandmasters. That algorithm values a delicate balance of factors such as pawn positions and the safety of its king.
On the other side was a new program called AlphaZero, a chess engine in some ways very much weaker than Stockfish – powering through just th as many moves per second as its opponent. But AlphaZero is a completely different machine. Instead of deducing the ‘best’ moves with an algorithm designed by outside experts, it learns strategy by itself through an artificial-intelligence technique that’s called machine learning.
Its programmers merely tuned it with the basic rules of chess and allowed it to play several million games against itself. As it learned, AlphaZero gradually pieced together its own strategy.
The head-to-head battle was astonishing. In 100 games, AlphaZero never lost. The AI engine won the match with dazzling sacrifices, risky moves, and a beautiful style that was completely new to the world of computer chess.
British chess grandmaster Matthew Sadler and mathematician and chess master Natasha Regan are still piecing together how AlphaZero’s strategy works in their new book Game Changer. We’re breaking open two moves in just one of the games to show the aggressive style, what it does, and what humans can learn from our new chess champion.
There’s a lot going on here, but focus on the pawns. Mainly, that AlphaZero has already lost one on the g file, and is sacrificing yet another with this jumpy rook move. (Stockfish’s next move is a queen leap to h2, gobbling up White’s lone soldier on the h file.) Run this position through many advanced chess engines, and most will tell you that, with the sacrificed pieces, AlphaZero is now losing. So why is it doing this?
Sacrifices are very common in chess, but they are almost always offered up for an immediate tactical edge or some other obvious recompense. But again and again, this magician-like chess engine makes early sacrifices such as these as part of an extremely long-term strategy whose benefit won’t become clear for dozens of moves into the future. Eventually, AlphaZero is going to fill the gaps left by the missing pawns with rooks, like a double-barrel shotgun. Those pawns, AlphaZero apparently believes, are worth less than the opportunity to assault the king from even more directions.
By move 42, AlphaZero has sacrificed even more pawns, and is marching another poor, disposable sucker towards oblivion. But this move seals AlphaZero’s victory. That final pawn is about to crack open Stockfish’s king’s corner like a knife twisting open an oyster.
Another key element to AlphaZero’s style is its absolute obsession towards attacks against the opponent’s king – rather than focusing on more delicate tactical plays.
By move 42, both of AlphaZero’s bishops control long open diagonals directed right at the king. Its queen is one leap away from the fray. And both rooks are likewise staring down Stockfish’s defence with unholy fury.
In their book, Sadler and Regan explain that it’s important for chess masters to embrace early strategic pawn sacrifices despite the risk: ‘Don’t rush! AlphaZero doesn’t attempt to deliver checkmate straight away but ensures that all its pieces are joining into the attack.’