Google's DeepMind AI team has once again beaten the world's best Go player at his own game.
Yesterday, the AlphaGo program beat South Korea's 9-dan Go player Lee Sedol in a live-streamed Go match. In the early hours of this morning, the AI program made it 2-0.
This time the game went to overtime, but Lee was eventually forced to resign. After the match, Lee revealed that it wasn't even as close as it appeared to be.
"Yesterday I was surprised, but today it's more than that – I am speechless," Lee told the press. "I admit that it was a very clear loss on my part. From the very beginning of the game I did not feel like there was a point that I was leading."
What was most notable about AlphaGo's victory, according to DeepMind head – and former game designer – Demis Hassabis, was that the program "played some quite surprising and beautiful moves."
Go's vastly high number of potential scenarios and strategic possibilities makes it a far bigger test for a computer program to master than, say, Chess. It requires the program to be able to evaluate and form strategies on the fly, rather than simply crunching through recorded scenarios.
Lee Sedol still has a chance of winning the match overall, but he'll need to win all three of the remaining games to claim the $1 million prize.
