Google’s AI system AlphaGo, part of its DeepMind project, has again beaten world champion Lee Sedol – and looks like it may be on track to take the title in the next game. Engadget reports Sedol saying that he was left speechless by his defeat.
“I’m quite speechless,” said Lee in the post-match conference. “It was a clear loss on my part. From the beginning there was no moment I thought I was leading.”
DeepMind CEO Demis Hassabis tweeted that AlphaGo “played some beautiful creative moves in this game” …
Sedol lost the first game yesterday, saying then that he was “very surprised.” He had predicted at the outset that he would win the five-game series 5-0 or 4-1 “at worst.” It’s now looking increasingly likely that it will be the human player, rather than AlphaGo, which fails to win a game
DeepMind said that more than 100M people worldwide watched the opening game online, 60M of them in China. There’s a $1M prize at stake, which Google says it will donate to charity if it wins.
The extent of the achievement by the DeepMind team cannot be overstated. While chess AI systems rely on number-crunching their way through all possible moves and counter-moves, AlphaGo takes a completely different approach, employing a neural net to learn for itself how to win the game. This is an approach truly deserving of the term artificial intelligence. Exciting times!