Become a fan of Slashdot on Facebook

 



Forgot your password?
typodupeerror
×
AI Google Games Technology

Google's AlphaGo AI Beats Lee Se-dol Again, Wins Go Series 4-1 (theverge.com) 111

An anonymous reader quotes an article at The Verge about Korean grandmaster's fifth and final game with Google's AlphaGo AI: After suffering its first defeat in the Google DeepMind Challenge Match on Sunday, the Go-playing AI AlphaGo has beaten world-class player Lee Se-dol for a fourth time to win the five-game series 4-1 overall. The final game proved to be a close one, with both sides fighting hard and going deep into overtime. The win came after a "bad mistake" made early in the game, according to DeepMind founder Demis Hassabis, leaving AlphaGo "trying hard to claw it back."
This discussion has been archived. No new comments can be posted.

Google's AlphaGo AI Beats Lee Se-dol Again, Wins Go Series 4-1

Comments Filter:
  • I imagine the next version will go 5-0 as these kind of things tend to be iterative in nature.

    • by Sique ( 173459 )
      Actually, even this version will be better at the next game, as it is a self learning system. It became that good at Go by playing millions of games against diverse opponents.
    • Re: (Score:3, Interesting)

      by Anonymous Coward

      Probably, but the fourth game was rather incredible and unlikely to be repeated again, even without changing alpha go. Lee took the corners and forced go into the much larger middle, Opposite of how he played game 2. This probably was the perfect setup to best alpha go, because it was too big an area to go deep into all possibilities of moves, yet to complex an area for the general strength of positions to be understood. They way the centre unfolded ended up a dream knife for Lee, Lee built to it when he sa

      • by umghhh ( 965931 )
        The estimations may vary of course. The opinions of experts is that the way the game 4 unfolded hit the weak spot of the machine. Maybe it is reproducible maybe not. A human had a chance to play 5 times against it and scored 1, the machine played zillion times and millions of games some of them from skilled humans - I see some disparity here. Another question is - all these hundreds of cores needed to best one of best humans at the game. The developers say that more cores would not do as diminishing return
        • the machine played zillion times and millions of games some of them from skilled humans - I see some disparity here.

          Yes, but a machine never forgets and can easily be replicated once it learns a task.

          Another question is - all these hundreds of cores needed to best one of best humans at the game.

          And the first computer took an entire room and could be beat by a guy with an abacus. Look where it is now.

          The end result is as with anything else - we produce a tool to replace our weaknesses - it was a shaped stone back in the cave, it is a 1.5k silicon cores now.
          If the machines ever get really intelligent the way we are - will they be modern slaves or will they succeed in freeing themselves.

          Luckily current computers are nowhere close to conscience which probably makes them better. Better to keep them as tools than a conscience entity that you have to worry about rights.

          From this perspective this result is irrelevant even if it were significant and it is less significant than people think it is.

          Not irrelevant at all. There are plenty of tasks that it would be nice to automate. Even something as simple as folding someone's laund

    • Black or white stepping stone?
    • by Anonymous Coward

      Lee Sedol may have the distinction of being the last human to ever win against a computer.

  • by Anonymous Coward

    and at that point go will be as bad as chess and it will be nigh impossible to find a fair game online.

    • by Anonymous Coward

      No. It will be running as GoAAS (Go As A Service) on The Cloud, while picking up your sexual orientation, location data and the scent of your underwear. Or something.

    • by ledow ( 319597 ) on Tuesday March 15, 2016 @05:51AM (#51698927) Homepage

      Most likely. But that's the case with just about any game.

      Can you really guarantee to find fair games of anything from tic-tac-toe, to chess, to draughts, to reversi, to Risk, to poker, to anything at all online? Go was pretty much out-there on its on in this regard but ordinary PC Go software has been able to beat amateurs for a few years now. This is orders-of-magnitude in terms of a leap in capability but nothing that would change the situation for the majority of people playing it.

      Pretty much the only "games" that can be fair are if you can guarantee it's against a human without any kind of possibility they could be plugging moves into a computer at any point. That's a vanishingly small amount of plays, pretty much limited to strict competitions (and even professional chess competitions have seen people use toilet breaks to illicitly get computer analysis of the board state on their phones!).

      What I want is not a computer player that never wins, nor one that wins all the times. Those are EASY to program in comparison to one that CONVINCINGLY challenges you enough that you have to play slightly better each time in order to win, without trouncing you or letting you walk all over it.

      That's the REAL hard problem in any kind of game "AI" - the "gamer's Turing Test" - how to lose/win convincingly without people knowing you're a bot.

      Try playing a pool game on a computer, for example. They usually go from "whoops, missed a blindingly obvious easy shot" to "four-cushion bounces, jump the ball, curve into the other balls coming back from the cushions, and tap one into the nominated pocket" without anything convincing in between.

      Like Left4Dead's "director" - we need to adjust to the player just enough to make it fun but that if they're obviously letting their guard down, we take advantage. Unlike the 80's arcade games that had to be punishingly hard but just easy enough at first for you to want to put money in but not to waste too much time in front of, modern video games need to be easy enough to pick up and get you into and keep you coming back for more (and spending some DLC) without feeling like you're playing a script, trouncing everything, or need to spend a fortune just to stay competitive.

      • >

        What I want is not a computer player that never wins, nor one that wins all the times. Those are EASY to program in comparison to one that CONVINCINGLY challenges you enough that you have to play slightly better each time in order to win, without trouncing you or letting you walk all over it.

        Amen. I happened to be trying to tune the Pachi [github.com] Go AI to something slightly better than my current level just last night. It's very frustrating -- one can control the number of cores and calculation time, and attempt to zero in from there, but each game takes long enough that (even on reduced-size boards) it's a slow process.

    • Comment removed based on user account deletion
      • No no, you're thinking is all wrong. In 10 years, we'll pitting our phones together on the same table and have them play it out while placing winning bets . It's sorta like putting two Furbies in front of each other; useless, but endless fun :)

        In 10 years, if you put two Furbies in front of each other, they'll spend a few minutes evolving a common private language, then agree to cooperate to kill you in your sleep.

        Elon Musk and Steven Hawking agree with me, so I know I'm right.

    • In 10 years this will run on phones.

      This had the power of 1024 CPUs and 250 GPUs. Even if CPU speed increases at twice the rate of doubling every two years (hint: that's not going to happen), we would not see this on the desktop in ten years. Google put a lot of processing power into this.

      and at that point go will be as bad as chess and it will be nigh impossible to find a fair game online.

      I have no problem getting a fair game online. I do it by being really bad. If someone is using a chess computer to win, then they will have a rating far higher than me :)

      • > we would not see this on the desktop in ten years.

        The 5d version was based in a single server. It only had something like 8GPUs and 40 CPU cores and still was pretty good.

        The 9d version is the distributed version that took so much processing power.

      • Even if CPU speed increases at twice the rate of doubling every two years (hint: that's not going to happen), we would not see this on the desktop in ten years

        Deep Blue needed a ton of hardware, including specialized VLSI chess chips, to narrowly beat Kasparov in 1997. Just 9 years later, World Champion Kramnik lost to a dual Xeon desktop PC.

        • Yeap, and deep blue was smaller than AlphaGo by two orders of magnitude (not include the clock speed increase we've seen since then, going based merely on CPUs). The GPUs in Alphago are doing essentially the same thing as the VLSI chips.
          • The Alphago computer may be 100 times as large, but that doesn't mean it's 100 times as fast. At first glance, games should be easy to parallelize, but that's not really true, as one node can find a good move that automatically invalidates a whole bunch of work that other nodes have already done. Also, transposition between move sequences (i.e. the same moves but in different order) means that work is duplicated on parallel systems. In addition, 10 years of further research will undoubtedly lead to better
            • n addition, 10 years of further research will undoubtedly lead to better software, just as was the case with chess computers.

              Maybe. DeepBlue's evaluation function was really lousy, but they made up for it with brute force. The further gains have been from improving the evaluation function.

              Compared to AlphaGo, whose evaluation function is already really good. We may see further improvement, but you can't generalize based on the experience with chess. It is possible we won't have a desktop-style AlphaGo for decades.

  • was that really an "early mistake" or was it part of the plan? how do we know?
    • I was wondering about that too. The funny thing is, because of how AlphaGo learns and plays moves, Google engineers cannot really tell either.

  • by Anonymous Coward

    I get tired of hearing people say that Go is a game that required creativity to win. It doesn't and if anything, this result demonstrates that.

    "AlphaGo's algorithm uses a combination of machine learning and tree search techniques, combined with extensive training, both from human and computer play."

    It's a game, based on pre-defined rules.It's just more opague and vague than chess.

    • Can you define "creativity" for us?

      • by Anonymous Coward

        My definition would be to do something novel. Doing something that nobody has done before isn't always novel. If I add two numbers together that have never been added together before, it's not really novel, you're just using a well defined method on a different data set. A lot of what people seem to be calling creativity with AlphaGo seems to fall into this category. Given a board state, it used a well defined algorithm and well trained statistical model to search through potential moves and identify wh

    • You clearly never played Go.

      It is not because you can 'solve' a problem by throwing some machine learning and tree search that this problem doesn't need some creativity to be solved by a human. The human has a very good machine learning capability, but I should say a very limited tree search capability. The human compensate that tree search capability deficiency by its very good machine learning capability and a touch of creativity. I dare you to do an extensive search like deepblue did 20 years ago with yo

  • If Google AI you can beat Lee Se-dol at Go, can it beat the IRS and Her Majesty's government at Tax Evasion? http://www.huffingtonpost.com/... [huffingtonpost.com] http://www.theverge.com/2016/1... [theverge.com] http://www.thelocal.it/2016021... [thelocal.it] http://www.bbc.com/news/magazi... [bbc.com] http://news.yahoo.com/italy-cl... [yahoo.com]
  • This was a great proof of concept for some "intuition" in AI, one of the behavioral aspects people believed hard to reproduce.

    Now I am really looking forward to see the real applications for this, and their consequences:
    - smart AI assistants, "a Siri that actually works" and similar
    - AI assisted science
    - AI assisted healthcare

    There is a great interview with Demis Hassabis [theverge.com] about this. There is hope for noticeable progress in mass products within 3-5 years.

    This new tech will help a lot of people dire

  • by javipas ( 1086007 ) on Tuesday March 15, 2016 @06:29AM (#51699035)
    I've been following the matches with the same expectation and anger I felt in 1997 during the Kasparov & Deep Blue rematch. The final result has been similar, and although it has been well reasoned that chess and go are pretty different games and Deep Blue and AlphaGo are pretty different machines, the bittersweet sensation is identical. I had a naive hope in the human superiority just for a little more time. I was pretty sad after the final game: Lee Sedol seemed really disappointed and sad himself. I can't imagine the pressure he's felt throughout the event, and his face -that's my impression- seemed to tell us "I've failed you all". He later told in the press conference that he felt he could have done more in the games -I'm sure he'd like to play more games to test himself again- and I wonder what could have happened if the matches would have been played without general knowledge. Feeling that kind of coverage must have been really stressful. If you ever read this, Mr. Sedol, thank you. And please, don't ever feel disappointed, you've done a fantastic job.
    • by Anonymous Coward

      Pfff...

      It's like being angry at a juicer because it makes juice better/faster than a human.
      Humans aren't out of the equation. The machine logic and innards are human produced.

      Thank you for doing this delicious juice, you've done a fantastic job. I'm grateful that people like you do the juice all by themselves, with their human hands!

      • Um, no. Obvious differences asides this is like being angry at a juicer because it writes better music than you.

    • I've been following the matches with the same expectation and anger I felt in 1997 during the Kasparov & Deep Blue rematch. The final result has been similar, and although it has been well reasoned that chess and go are pretty different games and Deep Blue and AlphaGo are pretty different machines, the bittersweet sensation is identical. I had a naive hope in the human superiority just for a little more time. I was pretty sad after the final game: Lee Sedol seemed really disappointed and sad himself. I can't imagine the pressure he's felt throughout the event, and his face -that's my impression- seemed to tell us "I've failed you all". He later told in the press conference that he felt he could have done more in the games -I'm sure he'd like to play more games to test himself again- and I wonder what could have happened if the matches would have been played without general knowledge. Feeling that kind of coverage must have been really stressful. If you ever read this, Mr. Sedol, thank you. And please, don't ever feel disappointed, you've done a fantastic job.

      He did seem visibly upset, as did Kasperov himself if I remember correctly. I don't blame him at all for losing - I think that he did an excellent job, and I agree with you that I'd love to see more. Ultimately though, we'll never know until we see it take on multiple would be champions, and maybe some rematches. I still don't think the computer has beat all of humanity yet - it hasn't had the history yet of being able to beat many people, and not just five or ten games, but consistently. Seeing it handle

    • I didn't felt the same this time as in 1997 because this time I was really betting on the computer to win. It is true that most professional didn't have imagined 1/10th of a second that the human will lose any match against AlphaGo due that the state of the art until now in Go AI were at most 1dan AI and it is hard to imagine a leap from 1dan to 9dan overnight. I really enjoyed seeing these games and thank Sedol for the great job he did under such circumstances. And if he think he failed me he is completely

    • by Anonymous Coward
      ... by saying "I Knew I shouldn't have had those two scotches before the game" and winking. Because when it comes to being a B*tard, humans can beat a computer every time!
    • Lee Sedol seemed really disappointed and sad himself. I can't imagine the pressure he's felt throughout the event, and his face -that's my impression- seemed to tell us "I've failed you all". He later told in the press conference that he felt he could have done more in the games -I'm sure he'd like to play more games to test himself again- and I wonder what could have happened if the matches would have been played without general knowledge.

      Yeah, I think playing against the unknown opponent really threw him off. Michael Redmond said in games 1 and 3 he used the wrong strategy for playing against the computer, and had he used a different strategy, his results would have been improved.

    • I feel the same way I felt after Watson won Jeopardy. I happened to be in my graduate AI class at the time and Watson was all the rage, subject to many class discussions. Every single one of us, teacher included, were really excited for what the future might hold.

      Does anyone remember that commercial with all the kids saying "I am Tiger Woods"? Today, just like at the time Watson triumphed, I like all others in my class at that time, feel like chanting with pride "I am an AI researcher!".

  • by Lisandro ( 799651 ) on Tuesday March 15, 2016 @09:16AM (#51699889)

    He's likely to be remembered as the last human being to beat a Go AI on tournaments.

    Move 78, in particular, was so good that his partners and commentators in China have already called it "the hand of God", but it really was one of those things which happens once in a blue moon, even for a player like Sedol.

    • I felt that both the commentaries on the official stream and the one of AYA with Kim Myungwan [xmp.net] were mostly focusing on analysing alternatives to moves that were already played, instead of focusing on possible moves to be played. At one point Haijin Lee [xmp.net], also known as Haylee, the current secretary of the International Go Federation, took a seat behind the computer. While Lee Sedol was taking a long time to think about the 78 move, she suggested the "wedge" move [youtube.com] to Kim Myungwan. He first asked with some surpis
  • I think it's obvious that computers will shortly be able to be any human player at virtually any kind of structured game. In fact, I have a hard time imaging a game where computers won't soon be able to beat a human.

    Even unstructured games like Pictionary and Cards Against Humanity will eventually be able to be played well by computers (after enough training and live competition). Determining the "winner" of those games is subjective, but I've little doubt that computers will eventually be able to master th

  • Now let's imagine : let two AlphaGo machines play each other Go games. More games. More time allowed... Folks : it becomes IMO so abysmal. Where will it stops ? I literally shiver in awe. I believe this could be radically extreme disruptive technology. Keep in mind, AlphaGo invented moves it never observed before. Keep in mind, it can learn quite some different games, just by being exposed to samples. Wooooooaaaaaa. Impressed, concerned, exited, I am. Z.
    • Now let's imagine : let two AlphaGo machines play each other Go games. More games. More time allowed... Folks : it becomes IMO so abysmal. Where will it stops ? I literally shiver in awe.

      Then this will blow your mind. This has already happened. AlphaGo trained against itself as a matter of course. In consequence, it has already played more games than any human alive ever could. Think about that for a while.

  • Go is an interesting game for this approach. It is thin, which mean that the moves and pieces are the same and don't do a lot. It is wide so calculating everything from scratch is essentially not doable...

    Chess got to be good enough by essentially matching GM search depth, by intelligently narrowing the search tree. And either capitalizing or avoiding tactical issues, within that depth, and if there are no tactical issues if there are a collection of moves that are left, make the ones that follow a di

  • " Thou shalt not make a machine in the likeness of a man's mind"- Orange Catholic Bible- Dune Series "The target of the Jihad was a machine-attitude as much as the machines," Leto said. "Humans had set those machines to usurp our sense of beauty, our necessary selfdom out of which we make living judgments. Naturally, the machines were destroyed."[6]-God Emperor of Dune "Man may not be replaced."-The Butlerian Jihad a.k.a. The Great Revolt use of technology trains humans to think like machines. The proble

The gent who wakes up and finds himself a success hasn't been asleep.

Working...