19:23 GMT +315 October 2019
Listen Live
    South Korean professional Go player Lee Sedol reviews the match himself after finishing the second match of the Google DeepMind Challenge Match against Google's artificial intelligence program, AlphaGo in Seoul, South Korea, Thursday, March 10, 2016

    Human Lands Blow on Supercomputer in Man vs Machine Board Game Showdown

    © AP Photo / Lee Jin-man
    Tech
    Get short URL
    1133
    Subscribe

    South Korean Go champion Lee Sedol, who has won 18 international titles playing the strategic board game, has managed to win one game against the AI program AlphaGo; he still trails by three games to one.

    A chair is pictured on stage as former US National Security Agency contractor Edward Snowden.
    © REUTERS / Svein Ove Ekornesvaag/NTB Scanpix
    Champion Go player Lee Sedol has won the fourth game of a five-game match against the AlphaGo computer program, having lost the previous three games to the supercomputer.

    "I've never been congratulated so much just because I won one game!" Sedol told a press conference after the game. 

    He had been predicted to win the series 5-0 or 4-1, despite AlphaGo already having beaten European champion Fan Hui in a five game match last October.

    Go is a strategic board game for two players which originated over 3,000 years ago in China, and is popular in many East Asian countries. It is called "Go" or "Igo" in Japan, "Baduk" in Korea and "Weiqi" in China.

    The game is played on a grid with black and white pieces, called stones, and the players score points for taking territory with their stones and capturing stones from the other player. 

    ​It was considered impossible for computers to play the game at a world-class level because the game has a near-infinite number of potential configurations and requires intuition to be played successfully.

    AlphaGo was developed by Google Deepmind, an artificial intelligence company founded in September 2010 by Demis Hassabis that was bought by Google in 2014.

    ​​Mistake was on move 79, but #AlphaGo only came to that realisation on around move 87

    ​The developers published a paper on AlphaGo in the journal Nature in January 2016, which explained the AI's success at playing humans, using "value networks" to select moves.

    They wrote that "these deep neural networks are trained by a novel combination of supervised learning from human expert games, and reinforcement learning from games of self-play," and enabled the computer to beat a human at Go, a feat previously thought to be at least a decade away. 

    Related:

    Game of Thrones? Tymoshenko ‘Fighting to Regain Power in Ukraine'
    Erdogan Plays 'a Bitter Game of Political Blackmail' With EU
    Banksy Hide-and-Seek Game Over
    Tags:
    machine, Artificial Intelligence (AI), game
    Community standardsDiscussion
    Comment via FacebookComment via Sputnik