Google's DeepMind AI Becomes a Superhuman Chess Player In a Few Hours

Discussion in 'BBS Hangout' started by No Worries, Dec 6, 2017.

  1. No Worries

    No Worries Contributing Member

    Joined:
    Jun 30, 1999
    Messages:
    18,262
    Likes Received:
    1,146
    Google's DeepMind AI Becomes a Superhuman Chess Player In a Few Hours

    In a new paper published this week, DeepMind describes how a descendant of the AI program that first conquered the board game Go has taught itself to play a number of other games at a superhuman level. After eight hours of self-play, the program bested the AI that first beat the human world Go champion; and after four hours of training, it beat the current world champion chess-playing program, Stockfish. Then for a victory lap, it trained for just two hours and polished off one of the world's best shogi-playing programs named Elmo (shogi being a Japanese version of chess that's played on a bigger board). One of the key advances here is that the new AI program, named AlphaZero, wasn't specifically designed to play any of these games. In each case, it was given some basic rules (like how knights move in chess, and so on) but was programmed with no other strategies or tactics. It simply got better by playing itself over and over again at an accelerated pace -- a method of training AI known as "reinforcement learning."
     
    Yung-T likes this.
  2. heypartner

    heypartner Contributing Member

    Joined:
    Oct 27, 1999
    Messages:
    49,565
    Likes Received:
    18,441
    This is also the method of training that @Jontro uses.
     
    B-Bob, LCAhmed, Invisible Fan and 2 others like this.
  3. ghettocheeze

    ghettocheeze Member

    Joined:
    Jun 9, 2006
    Messages:
    4,317
    Likes Received:
    2,377
    But can DeepMind score 90 in a half life the Rockets?
     
  4. Dr of Dunk

    Dr of Dunk Clutch Crew

    Joined:
    Aug 27, 1999
    Messages:
    30,722
    Likes Received:
    1,977
  5. CometsWin

    CometsWin Contributing Member

    Joined:
    May 15, 2000
    Messages:
    21,369
    Likes Received:
    2,872
    I'd like this computer to solve over-population and automation. See if it becomes self-aware.
     
  6. Air Langhi

    Air Langhi Contributing Member

    Joined:
    Aug 26, 2000
    Messages:
    17,639
    Likes Received:
    1,337
    It's a simple solution it's going to kill everyone.

    It seems like that is what unsupervised learning will do.
     
  7. HR Dept

    HR Dept Contributing Member

    Joined:
    May 31, 2012
    Messages:
    6,517
    Likes Received:
    907
    That computer doesn’t want it with me. I’ll open its browser and inundate it with tabs full of pop up laced porn sites and pirated streams. Wouldn’t know what hit it.
     
  8. Phillyrocket

    Phillyrocket Member

    Joined:
    Jun 12, 2002
    Messages:
    8,516
    Likes Received:
    1,159
    Reinforcement learning just sounds kind of ominous, Skynet like.
     
  9. A_3PO

    A_3PO Member

    Joined:
    Apr 29, 2006
    Messages:
    39,465
    Likes Received:
    1,972
    If/when artificial general intelligence happens, super intelligence will quickly follow.

    LOL, probably will take a week for it to figure out how to kill all mankind.
     
  10. RKREBORN

    RKREBORN Member

    Joined:
    Jul 24, 2006
    Messages:
    3,177
    Likes Received:
    563
    Can it have sex?
     
  11. RedRedemption

    RedRedemption Contributing Member

    Joined:
    Jul 21, 2009
    Messages:
    29,884
    Likes Received:
    3,902
    Yeah, but can it beat me in Starcraft?

    ...

    Oh ****.
     
  12. CCorn

    CCorn Contributing Member

    Joined:
    Dec 26, 2010
    Messages:
    11,478
    Likes Received:
    1,829
    Ai was more fun when it was morphed into a racist on twitter
     
  13. peleincubus

    peleincubus Member

    Joined:
    Oct 26, 2002
    Messages:
    15,518
    Likes Received:
    1,333
    I just do not see how this does not happen eventually.
     
  14. iconoclastic

    iconoclastic Member

    Joined:
    Oct 10, 2007
    Messages:
    6,036
    Likes Received:
    322
    I'll take it on in Clash Royale.
     
  15. sealclubber1016

    Joined:
    Mar 31, 2010
    Messages:
    9,433
    Likes Received:
    3,275
    So this is how it ends for the human race.

    We had a colorful run.
     
    MadMax likes this.
  16. txtony

    txtony Member

    Joined:
    Sep 18, 2008
    Messages:
    6,954
    Likes Received:
    1,107
    When is Morey adding Deepmind to the sideline?
     
  17. Rocket River

    Rocket River Member

    Joined:
    Oct 5, 1999
    Messages:
    42,487
    Likes Received:
    4,136
    If in the actual LIFE we live. . . . . your opponent is a machine.

    I look at the Robot Apocalypse and wonder . . . . could humanity lose to its creation.
    can you build something smarter than you?
    I think the biggest issue is building something more ruthless than you
    that is where we lose.

    Rocket River
     
  18. pirc1

    pirc1 Contributing Member

    Joined:
    Dec 9, 2002
    Messages:
    12,386
    Likes Received:
    525
    I think that is actually one of the project for the deepmind team, to train the AI to play video games like a human, with mouse and keyboard and using electronic eyes for data gathering on the screen. Most of the human jobs will be replaced in less than 50 years is my guess, maybe less than 20 years.
     
  19. VesceySux

    VesceySux Contributing Member

    Joined:
    Jul 31, 2001
    Messages:
    7,518
    Likes Received:
    135
  20. Rocket River

    Rocket River Member

    Joined:
    Oct 5, 1999
    Messages:
    42,487
    Likes Received:
    4,136
    What will we do with the excess humans then?

    Rocket River
     
Loading...

Share This Page