DeepMind AI teaches itself chess from scratch in four hours, proceeds to beat previous...

By William Gayde ยท 34 replies
Dec 9, 2017
Post New Reply
  1. DeepMind, Google's artificial intelligence focused cousin, has a history of creating computer systems that can defeat world champions at their own game. Back in 2016, they built an AI that beat a Go world champion. Since then, they have set their sights on chess and this week they pulled off another remarkable feat. Their game-playing AI known as AlphaZero has taught itself chess from scratch in just a few hours and then went on to beat the world's previous best chess-playing computer program.

    Unlike most computer problems, games like Go and chess don't have a clear solution. In chess, each turn can play out in any one of 400 different ways. This means that after just 4 moves apiece, there are nearly 300 billion different possible positions. A standard game has roughly 40 moves which makes the number of possible outcomes greater than the number of electrons in the observable universe.

    Clearly this approach of analyzing every possible move is not feasible, so that's where the artificial intelligence part comes in. AlphaZero was given only the rules of the game and no further human input. It then used a technique called self-reinforced knowledge to actually learn the details of the game.

    In this technique, the system plays itself over and over while learning along the way. They describe this as a "more human-like approach" to learning the game compared to traditional programs that run an algorithm.

    After all of this learning, the system was put head to head with Stockfish 8, one of the previous world champion programs. AlphaZero either won or drew all 100 matches played. The system isn't just designed for chess though. The designers also repurposed it to play Go against the original AlphaGo program. AlphaZero won that 60 to 40, too.

    Permalink to story.

     
  2. Uncle Al

    Uncle Al TS Evangelist Posts: 3,349   +1,991

    Hmmmmmm ....... so, we've taught a computer to play with itself? Commander Data would say that's' pretty close to being human!
     
  3. Kotters

    Kotters TS Maniac Posts: 283   +180

    How do you even get out of bed in the morning?
     
  4. Igrecman

    Igrecman TS Maniac Posts: 212   +123

    Most people thought a few decades ago that a lack of imagination and creativity would be the thing holding back computers in the future. Although chess seems to require a bit of creativity an imagination, it can be emulated by great processing power. However I believe an AI will not be able, in this century, to create by itself a best seller novel or a blockbuster CGI movie without human intervention. A machine would need to be more human for that to happen.
     
    Last edited: Dec 9, 2017
    Reehahs and fimbles like this.
  5. fimbles

    fimbles TS Evangelist Posts: 1,185   +208

    Chess is purely a game of skill with no room for chance. The only flawed component is the human player.
    The human noggin simply cannot compete with that level of processing power, I would not say emulated, I would say surpassed ! If you feed every known word in to a computer and tell it to arrange them in as many different ways as possible, One of the books it spits out will be don quixote (babels library) (Examples at bottom of page):). Skynet..Tinfoilhat..ect :)
     
    Last edited: Dec 9, 2017
  6. Skjorn

    Skjorn TS Addict Posts: 182   +114

    Yeahhh idgaf what science thinks, they don't know how many electrons are out there. I just laugh at that **** when they think they know it. Its all theory and will forever be theory.
     
  7. yRaz

    yRaz Nigerian Prince Posts: 2,320   +1,412

    I laugh at people who think they know more than the professionals. Exact number? No, but we have a pretty good idea
     
    dms96960, DaveBG and trgz like this.
  8. Vulcanproject

    Vulcanproject TS Maniac Posts: 218   +159

    Chess to me isn't creative, nothing like the sense of a novel or film is. Chess is a game and like any serious board game it has rules and probabilities that can be defined by mathematical algorithms. Anything that can be fairly easily defined in that way, a computer will eventually be better at than any human.

    It's far more difficult to define creative arts mathematically that's for damn sure. Unless of course you're talking about Hollywood who have a computer that churns out the next sequel or rehash quicker than anyone can hit the delete key, clearly.....

    The chances of AI reaching the sort of level required for creative thought this century are slim, but I think it is possible given time. In the end humans are mostly a product of their surroundings. Most stories and inspiration come from the lore of our civilization and personal experience gained over decades. A genuine AI will have to be capable of truly learning for itself in every breadth of field, growing from childhood collecting it's own perceptions of surroundings, and processing them in a feedback loop as thoughts.

    I suppose this kind of machine learning is a very primitive form of that on a very limited scope.
     
    fimbles, Reehahs, emmzo and 1 other person like this.
  9. Skjorn

    Skjorn TS Addict Posts: 182   +114

    It is still theory that the core of the earth is solid or not how tf they going to know about how many electrons are in the universe? I'm not saying I know more, just that they don't know that. They can guess all they want but it is all just theories.
     
  10. Manuel Diego

    Manuel Diego TS Booster Posts: 27   +26

    You really have no clue, lol. Stop embarassing yourself, or you will soon ask whether the earth is round or not (hey, maybe it's flat after all! I guess we will never know! :D).
     
    DaveBG likes this.
  11. yRaz

    yRaz Nigerian Prince Posts: 2,320   +1,412

    Theory =\= hypothesis

    You're really showing your lack of education
     
    DaveBG likes this.
  12. Skjorn

    Skjorn TS Addict Posts: 182   +114

    I'm not doubting that the center of the earth is solid. Just they don't know this for obserable fact which is different from theory. You don't even really have to look far into it to see that it is stated as theory. The article simply seemed like it stated that it was fact, when it is really just theory. Theory is really just a well agreed upon hypothesis that others have tested and come to the same results. It really isn't that hard to differentiate between these 3. Don't accept theory as observable facts, Show your education more, I know you can.
     
  13. yRaz

    yRaz Nigerian Prince Posts: 2,320   +1,412

    stay in school kids
     
  14. Skjorn

    Skjorn TS Addict Posts: 182   +114

    Ok, and just accept widely accepted guesses as facts right? Nice, an educated response. You should dig to center and just double check for them.
     
  15. Vulcanproject

    Vulcanproject TS Maniac Posts: 218   +159

    Using the word 'guess' is crude for a scientific theory based on an enormous amount of strong evidence refined over a long period of time to come to a general conclusion of a fact. You sound like a creationist talking about evolution.

    The beauty of the scientific method is that if strong evidence appears to support a contrary viewpoint then the viewpoint changes. But that evidence currently does not exist (for example the earth's core is really cheese) so you can be very confident of the current conclusions. Even if you could observe it and tell me, how do I know your eyes don't see cheese everywhere and you are delusional? Absolute certainty may not be possible, and I mean for anything. We usually make do with levels of certainty. That's for philosophy professors to discuss.

    Right about now a creationist would say but 'how do you KNOW the core isn't made out of cheese, since you can't directly observe it?' So a patient scientist with lots of better things to do would direct that person to the enormous raft of strong evidence to support his or her conclusion. That doesn't mean you can't challenge it, it just makes you look stupid if you do so without solid evidence to the contrary of your own.....
     
    Last edited: Dec 10, 2017
    Reehahs likes this.
  16. Skjorn

    Skjorn TS Addict Posts: 182   +114

    Right about now I would say there is also a hollow-core theory. Taking assumptions as fact is exactly the opposite of what drives science.
     
  17. yRaz

    yRaz Nigerian Prince Posts: 2,320   +1,412

    that's a CONSPIRACY theory
     
  18. Vulcanproject

    Vulcanproject TS Maniac Posts: 218   +159

    Not realizing that philosophers and linguists still argue about what a 'fact' really means is your weakness here. That's semantics.

    We take theories backed up by strong evidence to be much more certain than ones backed up with nothing. We take theories repeatedly tested and backed up by reams of strong evidence over long periods as fact. Does that mean they are now absolute certainties? Not really. It just means they are far more valid than any other theory, unless strong evidence to the contrary is discovered.

    You can't just fake it until you make it in science. Casting doubt on a strong theory is acceptable, as long as you back it up with very good evidence.
     
  19. Skjorn

    Skjorn TS Addict Posts: 182   +114

    You have any idea about the minds behind that theory?
    Not backed by nothing. Learn yourself.
     
  20. yRaz

    yRaz Nigerian Prince Posts: 2,320   +1,412

    the earth would not have a magnetic field if it did not have a molten iron core.
     
  21. Vulcanproject

    Vulcanproject TS Maniac Posts: 218   +159

    Backed by strong testable evidence? Or pseudoscience? This is the point of my posts. Show us the testable theory otherwise expect to get called out and ridiculed. My 'theory' that Jupiter is made of cotton candy is every bit as credible otherwise.
     
  22. Andromadus

    Andromadus TS Rookie

    What about Skynet?
     
  23. cliffordcooley

    cliffordcooley TS Guardian Fighter Posts: 9,728   +3,701

    Much like the electron count of the universe. We can't make a number large enough for our own sun. And that is not including the fact we don't have a number large enough for star count either.
     
  24. yRaz

    yRaz Nigerian Prince Posts: 2,320   +1,412

    well it's not a conspiracy theory and do have any idea how numbers work? We can make numbers as large as we want, we just add zeros. That's the beautiful part of Base 10
     
  25. cliffordcooley

    cliffordcooley TS Guardian Fighter Posts: 9,728   +3,701

    Yes I do know how numbers work. The problem with this de-lema is you think objects can be counted, based on whether they may or may not exist. You're trying to chastise someone for speaking the truth, and its not working.
     

Similar Topics

Add your comment to this article

You need to be a member to leave a comment. Join thousands of tech enthusiasts and participate.
TechSpot Account You may also...