DeepMind's Artificial Intelligence (AI) AlphaGo is the official winner in the Go match between man and machine.
The AI defeated South Korean champion Lee Se-dol for the third time in a five-game live Go match in the Four Seasons Hotel, Seoul. Two games remain but DeepMind's computer is the winner despite their results, with no chance for a comeback from Lee whatever points he gains in the remaining games.
AlphaGo beat the 33-year-old Go grandmaster in the first match on Wednesday and the second on Thursday, much to the shock of the global Go community and Korean commentators.
For the third game, Lee brought his top game and used all his time early. Using overtime, he kept on for 13 minutes.
The only option for Lee now is to save face by clinching at least one victory in the two remaining games to avoid a 5-0 humiliation. The Korean has 18 international titles and is a 9-dan, the highest rank possible in Go, and one of the top players in the world.
Lee expressed confidence before the games but has admitted shock after his losses and it is now doubtful he will clinch even one win against AlphaGo.
According to Kyunghyang Shinmun, following his second lost on Thursday, Lee met with four of his fellow professionals and spent the night listening to their advice and trying to formulate a plan to defeat AlphaGo.
They decided to play as many irregular moves as possible to use up the whole board to minimize predictable variables that the computer can guess, the reports said. The longer the game would last, the easier for the AI to calculate possibilities, more so than Lee, they concluded, and agreed to lower mistakes in the beginning and run a quick game. The meeting lasted until 6am on Friday.
The champion indeed rallied with a do-or-die attitude. Lee opened the game as black and used his turns quickly in the beginning, which seemed to indicate that he was playing with a set strategy in mind this time. He took a break after an hour of playing intensely with strong attacks.
Korean commentators said Lee seemed to be attempting to run a quick game by playing strongly at the start. However, after an hour into the game they said that the AI was answering the aggressive attacks without mistake, in an "iron wall" defense, which seemed to indicate that Lee's strategy was failing.
Two hours into the game commentators were leaning towards DeepMind's AI for the victory. The grandmaster was committed until the finish, showing considerable strain to concentrate and attempting end-game moves to find at least some chance to even the score.
The million dollar winning purse will now be donated to Unicef and STEM.
Lee is still paid $150,000 for playing all five games.
AlphaGo is two years old and not tailored to fight best against a particular opponent; rather to increase the probability of winning regardless of who it is going against, David Silver of the DeepMind team explained pre-match. Last year, it defeated a European champion 5-0, but this recent victory is the first won against a 9-dan professional, the highest skill-level for Go.
Go is a Chinese board game played by two players alternatively placing black and white playing pieces, with the goal to surround more territory than your opponent. Your opponent's stones can be removed by surrounding them with your own.
Google founder Sergei Brin made a surprise appearance at the Four Seasons Hotel for the third, crucial match. He met with Lee 30 minutes before the start of the game and went up to his room to view the game via broadcast.
DeepMind CEO Demis Hassabis is staying for the game. Alphabet executive chairman Eric Schmidt visited Seoul to watch the first match and praised the occasion as a "win for humanity."
Despite AlphaGo's victory, the fourth match and final match will still be played on Sunday and Tuesday, respectively, which will give a chance for the South Korean grandmaster to avoid total humiliation.