In game 4 of a best-of-five match between Google’s AlphaGo program and master Go player Lee Sedol, Lee has notched a victory.
The artificial intelligence program, which won the first three games, still wins the match. But Lee has ensured that it won’t be a clean sweep.
As we’ve reported, AlphaGo surprised many experts with its successful streak of games against Lee. Many had believed the game was too complex for a computer program to have mastered yet. But after three games, the computer had dominated the match.
But Lee surprised the computer midway through game 4, on move 78, with a placement choice that impressed the human commentators and threw off the A.I.
Then he clarified … “When I say ‘thought’ and ‘realisation’ I just mean the output of #AlphaGo value net. It was around 70% at move 79 and then dived on move 87.”
That is, the program had calculated its odds of winning higher than they actually were, because it hadn’t properly intepreted the significance of Lee’s move.
AlphaGo never recovered, and ultimately resigned the game.
After the game, a laughing Lee Sedol said, “It’s just one win, and I’ve never been congratulated so much just because I won one game.”
He said he was thinking back to the days before the match began, when he expected to beat the computer 5-0, or maybe 4-1.
“If today I was, let’s say, winning three consecutive games and if I had lost one single game it would have really hurt tremendously. But … because I lost three matches and I was able to get one single win, I think this one win is so valuable I would not exchange this with anything in the world,” he said.
The DeepMind team isn’t unhappy, either. The programmers say that they themselves aren’t good enough Go players to test AlphaGo’s weaknesses.
So, they said, when a master player like Lee defeats the program, it’s an opportunity to make the program stronger.