AlphaGo, Google’s DeepMind artificial intelligence program, has won the final game against Lee Sedol, the world champion for Go. AlphaGo has won the series 4 to 1.
Before the games Lee Sedol, a South Korean, was reported as saying “I’m honored to play against an AI invented by Google.” He seemed confident he could win the match.
The first three wins were taken by AlphaGo, Lee had won the fourth one. Members of the Go community are as stunned with the inventive, aggressive way AlphaGo won as the fact that it did at all.
The ancient Chinese board game was long considered impossible for computers to play at such a world class level because of the extreme need of intuition required to master its intricate strategies.
Go is a game of intense strategic placement and profound complexity, played on a 19 x 19 grid, where each player takes turns placing stones, black and white to surround the opponent’s territory. AlphaGo uses artificial neural network and reinforcement learning, both developments on classical AI techniques that already exist. Essentially, Go teaches itself to play.
Think of it like how Google Photos allows you to search for pictures with a dog in them because it holds the memory of numerous dog images that have been processed and broken down to the pixel level.
Demis Hassabis, the creator of AlphaGo describes Go as “a bit of a holy grail for AI research.” He feels that Lee’s defeat is not at all a loss for human intelligence. “Our hope is that in the long run we will be able to use these techniques for many other problems,” says Hassabis.