Turn on thread page Beta

Google AI beats world Go champion in 4 out of 5 games watch

    • TSR Community Team
    • Clearing and Applications Advisor
    • Thread Starter
    Offline

    21
    ReputationRep:
    TSR Community Team
    Clearing and Applications Advisor
    http://www.bbc.co.uk/news/technology-35761246

    Managed to catch the last 40 minutes or so of the game this morning. Very exciting.

    Lee Sedol seemed pretty shook up- his hands were shaking as he was preparing to resign.

    Lee previously said he was confident of winning, although he did revise this as the match approached.

    There's been a lot of speculation about how it would go over the past few months. I didn't think it would be a clear 5-0 win for Lee Sedol but I wasn't expecting AlphaGo to make such a strong start. I hope Lee Sedol doesn't take too much of a hit to his confidence.
    • Study Helper
    Offline

    14
    ReputationRep:
    Study Helper
    (Original post by Puddles the Monkey)
    I hope Lee Sedol doesn't take too much of a hit to his confidence.
    This :emo:

    I don't want it to be Deep Blue vs Kasparov all over again :moon:
    • TSR Community Team
    • Clearing and Applications Advisor
    • Thread Starter
    Offline

    21
    ReputationRep:
    TSR Community Team
    Clearing and Applications Advisor
    (Original post by Nirgilis)

    I don't want it to be Deep Blue vs Kasparov all over again :moon:
    It already is!
    • Study Helper
    Offline

    14
    ReputationRep:
    Study Helper
    (Original post by Puddles the Monkey)
    It already is!
    Heaven's above, have you tried to read the Nature paper describing this type of algorithm? :lolwut: . I wonder why Google decided to publish and put this in the public domain rather than protect its development... :holmes:

    It's always awkward when Google has more high-impact papers than your supervisor who's been researching at a University for 30 years :lol:
    • TSR Support Team
    • Clearing and Applications Advisor
    Offline

    21
    ReputationRep:
    TSR Support Team
    Clearing and Applications Advisor
    (Original post by Nirgilis)
    This :emo:

    I don't want it to be Deep Blue vs Kasparov all over again :moon:
    This is worse. Deep Blue's approach was primarily brute-force searching and took a supercomputer that was near the top 250th most powerful for its time. AlphaGo in comparison uses neural networks to act in a similar way to how a human thinks and learns and doesn't need as many resources.
    Offline

    21
    ReputationRep:
    This game is a game changer for ai


    Posted from TSR Mobile
    • TSR Community Team
    • Clearing and Applications Advisor
    • Thread Starter
    Offline

    21
    ReputationRep:
    TSR Community Team
    Clearing and Applications Advisor
    Holy :moo: Alpha Go won this morning's match too. Totally nailbiting game! It was a rollercoaster of emotions.

    It looks like it might be a straight 5-0 win for Alpha Go.
    • TSR Community Team
    • Clearing and Applications Advisor
    • Thread Starter
    Offline

    21
    ReputationRep:
    TSR Community Team
    Clearing and Applications Advisor
    There was this comment (paraphrased a bit) which I thought was chilling...

    "Korean commentators were shocked. They were trying to work out where Lee made a critical mistake but they were having trouble finding one.

    So either the mistake is beyond a human's comprehension, or Lee was fated to lose from the beginning."
    Offline

    20
    ReputationRep:
    Does this mean computers can solve scientific problems now?

    Can we get it to work on curing cancer? :beard:

    Or how about we get it to design an AI. Super computers build (super computers)^2. Future here we come
    Offline

    3
    ReputationRep:
    Skynet, incoming. Time to build a bunker.

    Posted from TSR Mobile
    • TSR Community Team
    • Clearing and Applications Advisor
    • Thread Starter
    Offline

    21
    ReputationRep:
    TSR Community Team
    Clearing and Applications Advisor
    (Original post by ChaoticButterfly)
    Does this mean computers can solve scientific problems now?

    Can we get it to work on curing cancer? :beard:
    I believe they're going to use the algorithm to solve other problems.

    I'm sad that Go was just a means to end to them :emo: I hope AlphaGo doesn't go away for ever.

    (Original post by DiddyDec)
    Skynet, incoming. Time to build a bunker.

    Posted from TSR Mobile
    :afraid:
    Offline

    20
    ReputationRep:
    (Original post by Puddles the Monkey)
    I believe they're going to use the algorithm to solve other problems.

    I'm sad that Go was just a means to end to them :emo: I hope AlphaGo doesn't go away for ever.



    :afraid:
    What happens if we mistreat computers though. Will that be a thing future left wing activists campaign about?

    Also computers can dream now :ahee:

    They don;t dream about electric sheep though :no: Rather terrifying goat devil dog things :-/
    Spoiler:
    Show
    • Community Assistant
    • Very Important Poster
    Offline

    20
    ReputationRep:
    Community Assistant
    Very Important Poster
    Interesting stuff. A step forward for AI .

    (Original post by Nirgilis)
    Heaven's above, have you tried to read the Nature paper describing this type of algorithm? :lolwut: . I wonder why Google decided to publish and put this in the public domain rather than protect its development... :holmes:

    It's always awkward when Google has more high-impact papers than your supervisor who's been researching at a University for 30 years :lol:
    I'm afraid it goes beyond my head. But it is interesting that they're willing to release it.
    • Study Helper
    Offline

    14
    ReputationRep:
    Study Helper
    (Original post by ChaoticButterfly)
    Does this mean computers can solve scientific problems now?
    Interestingly, it's possible that I'll be looking a little into machine learning and it's application into medical research as part of my PhD. So... yes! This type of research is underway :holmes:


    (Original post by Puddles the Monkey)
    There was this comment (paraphrased a bit) which I thought was chilling...

    "Korean commentators were shocked. They were trying to work out where Lee made a critical mistake but they were having trouble finding one.

    So either the mistake is beyond a human's comprehension, or Lee was fated to lose from the beginning."
    Considering the game is pretty much instinct, it doesn't surprise me that whatever 'mistake' was made we don't understand... What WOULD be interesting, is getting AlphaGo to re-play the game against itself, and 'suggest' the next best move for Lee given a certain position he was in, and then see how Lee's move compared to the computed best move...

    Hell, let's take it a step further and see what happens when AlphaGo plays AlphaGo


    (Original post by Kvothe the arcane)
    I'm afraid it goes beyond my head. But it is interesting that they're willing to release it.
    You and me both...
    • TSR Community Team
    • Clearing and Applications Advisor
    • Thread Starter
    Offline

    21
    ReputationRep:
    TSR Community Team
    Clearing and Applications Advisor
    (Original post by Nirgilis)
    Considering the game is pretty much instinct, it doesn't surprise me that whatever 'mistake' was made we don't understand... What WOULD be interesting, is getting AlphaGo to re-play the game against itself, and 'suggest' the next best move for Lee given a certain position he was in, and then see how Lee's move compared to the computed best move...
    It's not really all instinct... there's an element of it but I don't think that's quite right, especially at pro level where they are very very good at reading and calculating how much moves are worth. I keep seeing this instinct thing get mentioned in the press but I don't think it's quite the right way of thinking about Go, or at least it's a bit misleading. :beard:

    Hell, let's take it a step further and see what happens when AlphaGo plays AlphaGo
    AlphaGo has played AlphaGo hundreds of times - that's how it learns
    Offline

    20
    ReputationRep:
    (Original post by Nirgilis)
    Interestingly, it's possible that I'll be looking a little into machine learning and it's application into medical research as part of my PhD. So... yes! This type of research is underway :holmes:
    Does this means scientists have to worry about their job? :afraid:
    • TSR Community Team
    • Clearing and Applications Advisor
    • Thread Starter
    Offline

    21
    ReputationRep:
    TSR Community Team
    Clearing and Applications Advisor
    (Original post by ChaoticButterfly)
    Does this means scientists have to worry about their job? :afraid:
    Humans need not apply.

    We are becoming obsolete :yep:
    Offline

    20
    ReputationRep:
    (Original post by Puddles the Monkey)
    Humans need not apply.

    We are becoming obsolete :yep:
    Yeah I've seen that.

    Can;t come soon enough.

    We all live like Greek philosophers and upper classes off the back of our robot slaves.
    • TSR Community Team
    • Clearing and Applications Advisor
    • Thread Starter
    Offline

    21
    ReputationRep:
    TSR Community Team
    Clearing and Applications Advisor
    (Original post by ChaoticButterfly)
    Yeah I've seen that.

    Can;t come soon enough.
    It's exciting and terrifying. :yikes:
    Offline

    17
    ReputationRep:
    (Original post by Puddles the Monkey)
    AlphaGo has played AlphaGo hundreds of times - that's how it learns
    Yeah, millions, I think.

    This is really amazing. The difference in complexity between chess and go is astronomical. And the AI approach is totally different. Neural networks over brute force.

    The weird thing is, nobody knows how AlphaGo does it. Neural networks cannot really be probed to understand strategy or things like that. The computer can learn but cannot tell us how it learns.

    The scope for using AI to solve problems of science is amazing, and it really feels like we are on the brink. Some have speculated that we won't have professional scientists or diagnosticians in the future, just people who manage the AI computers!
 
 
 
The home of Results and Clearing

2,767

people online now

1,567,000

students helped last year
Poll
How are you feeling about GCSE results day?
Useful resources

The Student Room, Get Revising and Marked by Teachers are trading names of The Student Room Group Ltd.

Register Number: 04666380 (England and Wales), VAT No. 806 8067 22 Registered Office: International House, Queens Road, Brighton, BN1 3XE

Write a reply...
Reply
Hide
Reputation gems: You get these gems as you gain rep from other members for making good contributions and giving helpful advice.