An AI computer learned how to beat almost anyone at chess in 72 hours

We don’t stand a chance.
We don’t stand a chance.
Image: AP Photo/Adam Nadel
We may earn a commission from links on this page.

The computer’s domination over humanity may start at the chessboard.

Since IBM’s Deep Blue beat chess master Garry Kasparov in 1997—the first time a computer beat a reigning world champion at chess under tournament rules—computers have gotten faster and faster. But even the fastest modern computers are using the same technique Deep Blue employed—analyze every possible move as quickly as possible—whereas a human chess master might only consider a few moves at any given moment. A new artificial intelligence program, however, might have figured out how to let computers think more like we do when it comes to chess.

Matthew Lai, a computer scientist at University College London, recently published his master’s thesis, which demonstrated a machine learning system—called Giraffe, after this cartoon about evolution—that can learn to play at the International Master level of chess in just 72 hours. According to MIT Technology Review, Lai’s machine is a deep neural network—a computer system that’s inspired by the structure of the brain and attempts to learn and make decisions in a similar way. According to Lai’s paper, Giraffe performs “moderately better” than contemporary computer programs that analyze every possible move at once, as opposed to the few that might actually lead to success.

“Giraffe derives its playing strength not from being able to see very far ahead, but from being able to evaluate tricky positions accurately, and understanding complicated positional concepts that are intuitive to humans, but have been elusive to chess engines for a long time,” Lai said in his research paper.

Rather like many lonely humans, Giraffe learns by playing chess against itself, but as MIT Technology Review pointed out, it was also fed a massive dataset of moves from real chess matches. Lai used a database of millions of chess moves, adding in random other moves to build up a library of 175 million moves for Giraffe to call upon. Giraffe uses those moves in games it plays against itself, learning which worked in which situation, until it has enough of a knowledge base to take on other computer programs.

Giraffe apparently finds the best move from its top three move decisions for each round on 70% of moves. Lai found that after just 72 hours, Giraffe had the proficiency higher than about 98% of ranked human players.

When playing, Giraffe looks at the game in three ways, like a human might. First, it looks at whose turn it is and who has what pieces. Then it takes takes a holistic view of the entire board—what pieces are where. And finally it considers what moves each of its pieces can make. Based on the millions of moves it’s learned in the past (and whether they’ve had a positive, negative or neutral outcome), Giraffe will then make its move. With every new move, each new game, Giraffe learns a bit more and gets a bit better. Like humans do.

Right now, the system is not as fast as other engines, but Giraffe works smarter, not harder. Lai said he’s working to shorten the time it takes Giraffe to find the right move, and he’s confident he can get Giraffe up to Grand Master level pretty easily. “I still have many ideas that need to be explored,” he said.

Lai’s project shows that we’re getting to the point where we’re able to teach computers to complete tasks on their own, rather than have to program their every single action. For now, that might make computers better chess players, but it’ll also help create smarter self-driving cars, autonomous drones—and perhaps one day, real artificial intelligence.

Correction: This article originally said that Deep Blue’s victory over Kasparov was the first time a computer had beaten a human at chess under tournament rules, rather than the reigning world champion.