Google researchers in the UK have developed an artificial intelligence capable of beating top players at the game of Go, an abstract board game which originated in ancient China.
While machines have already bested humans at other games of strategy and intellect such as chess and Scrabble, Go remained a challenge for AI researchers. Some experts in the field believed that it would take a decade before a computer could defeat a pro at the game.
Google subsidiary DeepMind developed the software, called Alpha Go, based on a novel approach to machine learning which has proven superior to traditional strategies for creating winning AIs.
AlphaGo, successfully beat the three-time European Go champion Fan Hui 5-0 in a series of games at the company’s headquarters in King’s Cross last October. The results are only now being published in peer-reviewed journal Nature.
“When I lost the game, I was not happy, but all professionals will lose many games.” said Fan. “So I lose, I study the game, and maybe I change my game. I think it’s a good thing for the future.”
David Silver, DeepMind’s co-lead researcher said “AlphaGo looks ahead by playing out the rest of the game in its imagination, many times over.” The program involves two neural networks, software that mimics the structure of the human brain to aggregate very simple decisions into complex choices, running in parallel.
Researchers say that the same techniques employed by AlphaGo could be used to make advances in the field of robotics and other areas of scientific research. There could also be applications for improving digital assistants such as Siri and Cortana.
The researchers at DeepMind are now focused on taking down South Korean world champion Lee Se-dol. Lee has said that “regardless of the result, it will be a meaningful event in the baduk (the Korean name for Go) history. I heard Google DeepMind’s AI is surprisingly strong and getting stronger, but I am confident that I can win at least this time.”
By Eugenia Xiao
[Images via Gogame Guru // haly // Wbaduk]