Human 'Go' Champion Pulls Off Consolation Win Over Google's Computer

PHOTO: Go world champion Lee Sedol lost the first three games of the Google DeepMind Challenge but fought back to win the fourth one.PlayLee Jin-man/AP Photo
WATCH Human ‘Go’ Champ Pulls Off Consolation Win Against Google Computer

In the battle of man versus machine, humans scored an increasingly rare victory when Go Grandmaster Lee Sedol won a consolation round against Google's AlphaGo computer program.

Lee's victory came after he lost the first three games in a best-of-five tournament against Google's artificially intelligent program, showcasing one of the most remarkable advancements in AI since IBM's Deep Blue bested chess champion Gary Kasparov in 1997.

The development shows how stunningly smart computers are becoming but also shows the potential artificial intelligence can have in helping to solve real-world problems through everything from robotics to climate modelling and disease analysis.

Go, a board game that was played in ancient China, pits two players against each other. The players take turns placing black or white stones on a grid, with the object of dominating the board by surrounding the other player's pieces. The stones can't be moved unless they are surrounded or are captured by the other player.

While computers can now compete at the grand master level in chess, teaching a machine to win at Go has until now presented a unique challenge since the game has trillions of possible moves. It's estimated there are 10 to the power of 700 ways a game of Go could be played. By comparison, chess has around 10 to the power of 60 possibilities, according to Google.

While AlphaGo's three-peat victory is huge for the advancement of artificial intelligence, the loss to Lee also demonstrates that computers are not infallible.

Demis Hassabis, founder and CEO of Google's DeepMind, tweeted that Lee was able to pressure the computer into a mistake in game four, eventually leading to AlphaGo's loss.

The team was able to perfect AlphaGo by setting up two neural networks. One network predicts the next move while the other predicts the outcome. The more games it played against human experts and between its neural networks, the smarter the program has become, according to Google.

Google said in January its program won 99.8 percent of games against other programs designed to play Go -- giving it a nearly perfect record.

Facebook has also been working on solving Go through AI. CEO Mark Zuckerberg posted a congratulatory message on Facebook after AlphaGo won the series.

"Congrats to the Google DeepMind team on this historic milestone in AI research -- a third straight victory over Go grandmaster Lee Sedol," he wrote. "We live in exciting times."