Slashdot is powered by your submissions, so send in your scoop

 


Forgot your password?
Close
typodupeerror
×
AI Technology

Defeated Chess Champ Garry Kasparov Has Made Peace With AI (wired.com) 106

Posted by msmash from the closer-look dept.
Last week, Garry Kasparov, perhaps the greatest chess player in history, returned to the scene of his famous IBM supercomputer Deep Blue defeat -- the ballroom of a New York hotel -- for a debate with AI experts organized by the Association for the Advancement of Artificial Intelligence. He met with WIRED senior writer Will Knight there to discuss chess, AI, and a strategy for staying a step ahead of machines. From the report: WIRED: What was it like to return to the venue where you lost to Deep Blue?
Garry Kasparov: I've made my peace with it. At the end of the day, the match was not a curse but a blessing, because I was a part of something very important. Twenty-two years ago, I would have thought differently. But things happen. We all make mistakes. We lose. What's important is how we deal with our mistakes, with negative experience. 1997 was an unpleasant experience, but it helped me understand the future of human-machine collaboration. We thought we were unbeatable, at chess, Go, shogi. All these games, they have been gradually pushed to the side [by increasingly powerful AI programs]. But it doesn't mean that life is over. We have to find out how we can turn it to our advantage. I always say I was the first knowledge worker whose job was threatened by a machine. But that helps me to communicate a message back to the public. Because, you know, nobody can suspect me of being pro-computers.

What message do you want to give people about the impact of AI?
I think it's important that people recognize the element of inevitability. When I hear outcry that AI is rushing in and destroying our lives, that it's so fast, I say no, no, it's too slow. Every technology destroys jobs before creating jobs. When you look at the statistics, only 4 percent of jobs in the US require human creativity. That means 96 percent of jobs, I call them zombie jobs. They're dead, they just don't know it. For several decades we have been training people to act like computers, and now we are complaining that these jobs are in danger. Of course they are. We have to look for opportunities to create jobs that will emphasize our strengths. Technology is the main reason why so many of us are still alive to complain about technology. It's a coin with two sides. I think it's important that, instead of complaining, we look at how we can move forward faster. When these jobs start disappearing, we need new industries, we need to build foundations that will help. Maybe it's universal basic income, but we need to create a financial cushion for those who are left behind. Right now it's a very defensive reaction, whether it comes from the general public or from big CEOs who are looking at AI and saying it can improve the bottom line but it's a black box. I think it's we still struggling to understand how AI will fit in. Further reading: Fast-and-Loose Culture of Esports is Upending Once Staid World of Chess; and Kramnik and AlphaZero: How To Rethink Chess.
This discussion has been archived. No new comments can be posted.

Defeated Chess Champ Garry Kasparov Has Made Peace With AI More

Defeated Chess Champ Garry Kasparov Has Made Peace With AI

Comments Filter:

    • Re: (Score:1)

      by Anonymous Coward

      Google found this for me: https://www.mckinsey.com/busin... [mckinsey.com]

      • Which doesn't say much. I think this is a very low estimate, solely based on the job title. But creativity is needed in many more jobs than e.g. artists or designers.

        And you could even argue that creativity might be overblown and we might see AI have *some sort of* creativity that will have some value and therefore partially replace (or amplify) the so-called creative jobs.

        I agree with Kasparov though, when he says "I think it's we still struggling to understand how AI will fit in.". Indeed, this is all v

      • That report doesn't say 4% of jobs require creativity. It says 4% of work activivies require high levels of creativity, and 29% of work activities require median levels of creativity.

        So 100% of workers could spend 4% of their time being highly creative, which is totally different from 4% of jobs requiring any creativity.

        I seems that Kasparov heard of the McKinsey report but has a very muddled understanding of what it is actually saying.

        So:

        1. Most human jobs likely require some level of creativity.

        2.

    • Re:"4 percent of jobs in the US require creativity (Score:4, Interesting)

      by ranton ( 36917 ) on Friday February 21, 2020 @01:51PM (#59751558)

      Wasn't a tough Google search:

      Capabilities such as creativity and sensing emotions are core to the human experience and also difficult to automate. The amount of time that workers spend on activities requiring these capabilities, though, appears to be surprisingly low. Just 4 percent of the work activities across the US economy require creativity at a median human level of performance. Similarly, only 29 percent of work activities require a median human level of performance in sensing emotion.

      source [mckinsey.com]

      • That's about the dumbest thing imaginable. So construction workers don't have to use creativity? Farmers? Asbestos inspectors? Hair stylist? None of those things require creativity.

        For gawd sakes. It's hard to think of a job that doesn't require significant creativity.

        • Re: (Score:3)

          by Kjella ( 173770 )

          I want a building that's built to code. Food that's safe to eat. Comprehensive inspections. If any of those "get creative" it's usually a bad thing. And my hair stylist should only get creative if I ask them to. Yeah, they might have to resolve practical problems that come up. But if you've seen a modern farm you'll realize that your carrots go from seeds to bagged produce without ever [youtube.com] being near human hands. All we're talking about here is taking it to the next level, instead of sending out a guy with a tr

          • You have either no idea what creativity is, or you have never worked a craftmanship job in your live.

            Creativity is not only about determining the aesthetics of the final product. It's about solving problems efficiently, big or small. it's the same talent that makes a songwriter write a good song which also makes a Mc Donald's guy figure out where to put the trash dumpsters on a windy day so that smell the street table customers.

            • Creativity might solve a novel problem quickly, but it is not how to solve problems efficiently. That's just silly. You're confusing Virtue with Success.

              Solving problems efficiently requires knowledge of the problem domain, good tools, and troubleshooting methods.

              Getting creative in fixing a farm implement gets your hand chopped off.

              • [quote]
                requires domain knowledge...
                [/quote] ...which a person working in that respective domain certainly has. My point is that domain knowledge is not enough (just like, admittedly, creativity alone is not enough, but nobody claimed it was; I just claimed that it was necessary).

              • To stick to your farm tool example: it doesn't magically tell you what the problem is, you see only the symptom. So if I have an advanced broken farm tool (or a car, or whatever) I would certainly like l someone who's creative enough to find and fix the root cause, not just proficiently replace one expensive part after the other while the problem keeps popping up.

                See, most problems are not only about the problem domain itself, but also about their embedding in the rest of reality. Whoever works a craft know

          • You obviously have no idea what any of that means.

            Enjoy your stupid boring dumb life.

            • Re: (Score:2)

              by Kjella ( 173770 )

              Ah, hubris... have you seen that "Go away or I will replace you with a very small shell script" t-shirt? While it's humor there's also plenty tales of actual data entry/aggregation/consolidation workers put out of a job. The 2.0 version of that shirt is "Go away or I will replace you with a very small neural net". For example many factory line QA jobs have already been eliminated by computer vision, if that doesn't look like a normal input/output discard it and throw an alert if the error rate exceeds thres

              • Factory line QA jobs are NOT 4% of all jobs.

                I'm not saying that computers and machines can't replace a lot of human work, especially the repetitive work. But there is a long long distance between saying Computers can replace many repetitive and/or algorithmic tasks and "Only 4% of jobs require any creativity at all".

                Look at it this way, A company like Google can probably successfully develop a system that can replace professional drivers. But that is honestly the most repetitive/algorithmic profession of t

        • Yeah, construction workers don't really use creativity.

          • Said someone that clearly doesn't work in construction casting judgement on those that do.

            • Construction plans tell workers what to build. They don't tell the construction workers HOW to build it.

              This isn't IKEA.

              • Construction plans tell workers what to build. They don't tell the construction workers HOW to build it.

                Oh, the plans often include exactly how something is supposed to be built. It doesn't mean construction workers read it, or respect the engineer.

                And oh how they whine and cry when they have to rebuild it, "because of those darn pencil-necks with their clipboards!"

      • Re: (Score:1)

        by Anonymous Coward
        They're seriously misunderstanding creativity. Any slight deviation from the norm in a job requires creativity to handle the deviation. Creative solutions happen every day in nearly every job. Without creativity, the optimal solution is "shut down until environment is conducive to work", and that's a sure path to profit loss.

      • Re: (Score:3)

        by Toonol ( 1057698 )
        "4% of work activities" is a lot different than "4% of jobs". It could be an average of 4% of every job requires creativity... which would be closer to the truth.

        • I imagine that would even be a low estimate.

          Lets just think about mowing a lawn, for example. A computer that doesn't know what they it's doing just pulls out the lawn mower and shreds everything in front of it. A HUMAN that does know what they are doing quickly looks at the lawn,, assesses the scope of what needs cutting, decides on the fastest way to complete the job to the satisfaction of the homeowner and their boss, and then executes. Then you go to the next lawn which is 100% entirely different and do

  • .... players, a computer must evaluate millions of permutations, analyzing every possible move to find a winning sequence.

    A chess grandmaster, on the other hand. considers but the tiniest fraction of all these possible moves, and still plays objectively strongly.

    Why?

    What is the grandmaster doing that no computer yet made can?

    • .... players, a computer must evaluate millions of permutations, analyzing every possible move to find a winning sequence.

      A chess grandmaster, on the other hand. considers but the tiniest fraction of all these possible moves, and still plays objectively strongly.

      Why?

      What is the grandmaster doing that no computer yet made can?

      They're expecting their opponent to play in a similar fashion. They only have to think deep down the tree along the branches they expect their opponent to consider.

    • Nothing. That is why the computers win at chess 100% of the time now.

      • Re: (Score:2)

        by mark-t ( 151149 )
        Untrue.... a grandmaster can beat another competive player considering but the tiniest fraction of the number of board combinations a computer would have to evaluate to achieve the same ends.
        • No computer program is made to play chess that way because it's a stupid way to play chess when you have the ability to.

        • Untrue.... a grandmaster can beat another competive player considering but the tiniest fraction of the number of board combinations a computer would have to evaluate to achieve the same ends.

          This is true for classic chess programs like Stockfish.

          It is not true for AI-based chess programs like Alpha-Zero.

          Stockfish uses a deep brute force search.

          Alpha-Zero's strategy is more akin to human intuition. It is just much better at it than we are.

    • Pattern recognition. Human brains are really good at it, while computers suck at it. It's even a bigger factor in Go, which is why it took longer for computers to get good at it.

      • Pattern recognition. Human brains are really good at it, while computers suck at it.

        LOL do you even computer, Bro?

    • Re:In order to beat even very moderately proficien (Score:5, Insightful)

      by frank_adrian314159 ( 469671 ) on Friday February 21, 2020 @02:15PM (#59751656) Homepage

      What is the grandmaster doing that no computer yet made can?

      Selective pruning driven by their perception of best strategy. Sadly, this pruning and mistakes in finding the best strategy also means that errors can be introduced. In the past a grandmaster could recover against a human opponent (who also made these mistakes), but against a machine, there's little room for this kind of error. Which is why machines can now generally win against even grandmasters.

      • Re: (Score:2)

        by mark-t ( 151149 )
        I don't think you understand what I am asking.

        When playing another human, a grandmaster might consisder perhaps a few dozen or so maybe even as many as a couple of hundred different moves before deciding on one when playing against a human who plays at a competitive level.

        A computer, to achieve the same ends, must evaluate many orders off magnitude more board positions and moves than the grandmaster.

        Yes, it's obvious that the grandmaster is doing selective pruning and this does introduce the possibili

        • Why can't a computer emulate pruning as well as a grandmaster does

          It can, but you wouldn't want it to. You'd limit it to the best a human can be, when computers already do better than the best humans. You can't make a computer do better by using the human strategy, because the computer is already doing better due to the weakness in the human strategy. This is part of Kasparov's message: learn what human strengths are, and what computer strengths are.

          Humans aren't playing the game in the same way computers are. Oh, we do use the metrics, try to do the points math for the p

        • People thought AI wouldn't beat humans at Go because even computers can't explore any significant fraction of future board states in Go. But now the AI's use neural nets to generalize over past experience and "subjectively" evaluate board states, somewhat analogous to humans. So it would be interesting to see how few such "subjective" evaluations a computer would need to beat the best humans. Perhaps this has been done already, I don't know.

        • A computer, to achieve the same ends, must evaluate many orders off magnitude more board positions and moves than the grandmaster.

          No, it does not; the modern AI based ones like alphaZero considers just a tiny options. They work/think exactly like a human. Just that they could be made to train over a night using millions of servers; while a chess prodigy needs to spend 2 decades (of night sleep) to process n arrive at the same network.

    • Re: (Score:1)

      by dvice ( 6309704 )

      > What is the grandmaster doing that no computer yet made can?

      I don't understand your question. How do you think AlphaZero works? It doesn't even need to think, it can just make the move at the moment it sees the chessboard. And unlike chess grandmaster, it has beater the best chess computer that actually evaluates these millions of permutations. Oh and it took only a couple of hours for it to learn how to play chess from zero, without any external help to the level where nothing can beat it.

      • What exactly do you think AlphaZero is doing with all those GPUs?
        • AlphaZero doesn't use the GPUs in the way you're thinking.
          AlphaZero's chess inference engine uses TPUs (Tensor Processing Units)
          It's a neural net- it does not evaluate every possible permutation. It's net was trained (by itself) to play Chess.
          • OK, what exactly do you think AlphaZero is doing with all that processing power?
            • What is a chess grandmaster doing with all his neurons? Not evaluating billions of separate moves, but doing a more complex evaluation of each one.

              Also, are you talking about during training or during play? AlphaZero can learn from scratch to beat the best-ever human players in like 2 days. So that requires some processing. Not sure how much of that it really needs to play 1 human interactively.

            • Don't understand what you're trying to say.
              What's it doing with all that processing power? It's running a big ass neural network that has been trained to be good at Chess.
              What's a human doing with all that processing power? Certainly not calculating every possible permutation of the board.
              • AlphaZero is a tree searching algorithm at core. It uses the GPUs to search through a LOT of positions.
                • At it's core, sure, it is.
                  But the trick is that the neural net can infer the best possible moves, every move without having to evaluate all the permutations, i.e., intuition.
                  A lot like a human would.

                  I.e., is the MCST with a big neural network really different from what a human does?

                  • But the trick is that the neural net can infer the best possible moves, every move without having to evaluate all the permutations, i.e., intuition.

                    A) It doesn't find OR infer the best possible moves,
                    B) Computers have been playing without evaluating all the permutations since at least the 70s,
                    C) AlphaZero is somewhat better at tree-pruning than other computers, that's why it beats them.
                    D) AlphaZero is horrible at tree-pruning algorithms compared to humans, the only way it wins is by searching through millions of positions.
                    E) Humans probably don't have a variation of MC in their chess-playing algorithm.

                    • A) It doesn't find OR infer the best possible moves,

                      Actually, that's literally what the neural net does. Period. With increasing accuracy the more it is trained. These moves are then fed into a MC algo for evaluation.

                      Computers have been playing without evaluating all the permutations since at least the 70s,

                      True, but only kind of. They used a very simple pruning algorithm, and then used as much time as they had to test based on hard-coded weighting.

                      C) AlphaZero is somewhat better at tree-pruning than other computers, that's why it beats them.

                      No, it has its neural network, which is better by many orders of magnitude. [sciencemag.org]

                      D) AlphaZero is horrible at tree-pruning algorithms compared to humans, the only way it wins is by searching through millions of positions.

                      AlphaZero's pre-pruning *network* (stop calling it an algorithm) is definitely probably not as good as a humans (or maybe it is

                    • In summary, AlphaZero sucks at pruning compared to humans.

                      We can also add that AlphaZero was given extra processing power when it played against Stockfish, and if it played with similar hardware, it would have lost to Stockfish.

                    • In summary, AlphaZero sucks at pruning compared to humans.

                      You seem oddly invested in the above conclusion. Are you feeling insecure?

                      We can also add that AlphaZero was given extra processing power when it played against Stockfish, and if it played with similar hardware, it would have lost to Stockfish.

                      From the paper:

                      Stockfish and Elmo used 44 central processing unit (CPU) cores (as in the TCEC world championship), whereas AlphaZero and AlphaGo Zero used a single machine with four first-generation TPUs and 44 CPU cores

                      AlphaZero searches just 60,000 positions per second in chess and shogi, compared with 60 million for Stockfish and 25 million for Elmo (table S4). AlphaZero may compensate for the lower number of evaluations by using its deep neural network to focus much more selectively on the most promising variations (Fig. 4 provides an example from the match against Stockfish)—arguably a more humanlike approach to searching, as originally proposed by Shannon (30). AlphaZero also defeated Stockfish when given 1/10 as much thinking time as its opponent (i.e., searching ~1/10,000 as many positions) and won 46% of games against Elmo when given 1/100 as much time (i.e., searching ~1/40,000 as many positions) (Fig. 2). The high performance of AlphaZero with the use of MCTS calls into question the widely held belief (31, 32) that alpha-beta search is inherently superior in these domains.

                      You're being defensive about something from a completely uninformed position.
                      You can't even compare AlphaZero and Stockfish.
                      And most significantly, Stockfish uses algorithms programmed from observing real players- it uses known human moves.
                      AlphaZero taught itself completely from scratch. The only games that went into training its neural net were games it played against itself.
                      It literally taught itself to play Chess

                    • Hardware imbalance: "Stockfish and Elmo used 44 central processing unit (CPU) cores (as in the TCEC world championship), whereas AlphaZero and AlphaGo Zero used a single machine with four first-generation TPUs and 44 CPU cores"
                    • Yes, it is an imbalance, because AlphaZero uses a neural network, which runs on specialized Tensor cores (TPUs)
                      This is offset by the fact that AlphaZero's CPUs are practically idle in comparison (The units running the search tree)
                      This is on account that the search tree is fed by the neural network a highly pruned and inferred move list, including rankings of every move. This allows the CPU cores (running the search tree) to avoid having to do full monte-carlo rollouts.
                      It's why, in spite of having a *mass

                    • In summary... they can't have hardware parity.

                      TCEC manages it. Google just isn't trying.

                    • TCEC manages it. Google just isn't trying.

                      What kind of stupid fucking comment is this?

                      TCEC has no rules for TPUs. It doesn't manage shit.

                      AlphaZero can't enter into TCEC until the TCEC makes rules for TPUs. You can't efficiently run neural nets on standard scalar processors.
                      Saying Google isn't trying is like claiming the only reason Ferrari doesn't mop up at go-cart tracks is because it's not trying.

    • .... players, a computer must evaluate millions of permutations, analyzing every possible move to find a winning sequence. A chess grandmaster, on the other hand. considers but the tiniest fraction of all these possible moves, and still plays objectively strongly. What is the grandmaster doing that no computer yet made can?

      Relying on experience and pattern recognition. It's when they're forced to do deep calculations that GMs make most of their mistakes.

      • Re: (Score:2)

        by mark-t ( 151149 )

        I'm not disputing this.... but a GM can still beat other players considering only a fraction of the moves that a computer would to achieve the same result. If we make a computer only consider the number of moves that a GM actually considers, then we end up with a player that even a modestly competent player could beat.

        Why?

        • Two different ways of arriving at a move. The GM uses his experience to look at only a small handful of candidate moves. It is for this reason that even top GMs overlook strong moves that don't fit into their known patterns. It is not unusual for a GM, when shown a winning move they overlooked, to admit that they didn't even give it serious consideration.

          Consider this position: White to win [twitter.com]

          The computer will quickly see that Rf5 wins on the spot. Most human players would not even look at that move becaus

          • Re: (Score:2)

            by mark-t ( 151149 )
            So one is left then with the question, why do so many board positions "fit" a GM's experience? Why can't a computer play as strongly (while also just as fallibly) with just as little information?

            How do you encode years of play into an intuitive understanding of the game such that you don't *have* to consider every possible outcome just to arrive at a move that still has the best chance of winning against a player of comparable but still slightly less skill?

            • You want me to explain how the human brain works? I can't do that.
              You say "Why can't a computer play as strongly (while also just as fallibly) with just as little information?" How much information is that? How do you measure it? It might be a lot. The traditional computer approach to chess is to do a comprehensive search coupled with a clever position evaluator. Alpha Zero does not do that. Perhaps it is already doing what you ask.

        • Chess computers use all sorts of clever algorithms that take advantage of their capabilities, that is why.

          The computer doesn't decide what to calculate, the software developer decides what the computer evaluates.

          It is humans all the way down. The computer hardware was designed by a human engineer. The computer has no discrete self, no independent origin, it is entirely a time-saving device that makes calculations that humans can think of how to do. A computer cannot be good at things that humans are not goo

    • I'm not (much of) a chess player so take this with a grain of salt, but don't chess players memorize board positions? It's obviously more apparent with the starting moves, but there is a "best" (or series of best) line of play given the board, so you don't need to evaluate all options, you already have your tree pruned for you because it's been memorized.

      Later in the game, it's probably similar as pattern recognition helps your prune those branches very early. You think grandmasters haven't considered all

      • Re: (Score:2)

        by mark-t ( 151149 )

        The computer could do the same thing

        And yet, when a computer does just that, only considering about as many moves as the best human players alive, it becomes so fallible that it can be beaten by even an above average player, and is no challenge at all for a grandmaster.

        • Really? I'm not trying to be snarky. It's not only limiting the depth of your tree search, it's having to store thousands of board permutations and the relevant tree structure. Do the computers you describe do that, or do they only prune the tree/limit the depth of search?

        • Re: (Score:2)

          by dryeo ( 100693 )

          There's been chess games written that had a library of games to reference, so in the early game, rather then considering thousands of moves, looks up classical defenses or attacks and tries to follow them. Worked well in the early game against book learned chess players, not so well in the late game where calculation became more important.
          Generally this what master chess players do, follow certain memorized strategies until the end game..

    • AlphaZero does not compute every possible move. It's a large, complex, hardware assisted neural net running on TPU units.
      AlphaZero plays like a human. A human who has a better handle of Chess than any human.
    • Human also does millions of calculations -- but it is spread over say 10, 20 years. He builds his backend/brain rnn to recognize patterns. Today's AI computer chess like alphaZero works similarly. They take a tiny amount of time to find the move just like humans (similar to how they recognize say the ten digits, 0 to 9 in image recognition). But like humans have to train their network using trillions of computations (thru' things like self-play).
      The deep blue in late 90s, did not do AI; they were brute fo

    • Re: (Score:2)

      by nagora ( 177841 )

      You are buying into the same lie that Kasparov and many others did: that a computer can play chess.

      This like saying a motorbike can run. It can do the basic thing - move from here to there. But whatever it's doing, it's not running and a computer isn't playing a game.

  • I read somewhere, when I was a teenager, that there were people that were paid for what they did and there were people that were paid for what they know.

    Oh sure, a cocktail waitress has to know some things, and a brain surgeon has to do some things, but the difference is clear.

    I made the decision right then on which I would rather be, and it was the right one.

    • Yeah, I'm pretty happy with my tips, too.

    • Re: (Score:3)

      by vux984 ( 928602 )

      "Oh sure, a cocktail waitress has to know some things, and a brain surgeon has to do some things, but the difference is clear."

      Is it ? I'm not sure it's clear at all. In fact, I suspect we'll still have cocktail waitresses long after brain surgery is automated, which I suspect is not what you were trying to say.

    • The difference is actually not so clear.
      As the most important thing about a brain surgeon is his dexterity and ability to where he wants to cut.
      Where he has to cut is probably determined by some other doctor with fancy instruments.

      So, yes the surgeons bottom line "know more" ... but if you want to compare that with a bar tender who knows 10,000 different mixes then this "special know more" gets pretty mood.

    • Re: (Score:2)

      by dryeo ( 100693 )

      Yet the brain surgeon, a glorified barber, gets paid way more then the cocktail waitress who has to know how to deal with all kinds of weird people.
      It shows that by making a job more exclusive, in the case of surgeons by making them take a medical degree first and be at the top of their class, they can charge more money.

  • Knowledge worker (Score:1)

    by Anonymous Coward

    the first knowledge worker whose job was threatened by a machine

    What about the 'human' calculators from early NASA days?

    • Re: (Score:2)

      by dryeo ( 100693 )

      the first knowledge worker whose job was threatened by a machine

      What about the 'human' calculators from early NASA days?

      Clerks throughout the ages. Banking used to be done with pencil and paper, whole rooms of people computing balances and such. Mostly simple arithmetic rather then advanced math but sciences like astronomy also heavily depended on human computers and had practical uses such as almanacs for ship navigators to pinpoint their position in the middle of the ocean with a watch and sextant.

  • So are Go, Shogi, etc. Deterministic means if you play the same order of moves, the same result occurs every time. There's no randomness, no variation, no uncertainty. The current AI we have excels in situations with no randomness, no variation, no uncertainty.

    AI which can sort of deal with randomness, variation, and uncertainty is still in relative infancy, and still makes glaring mistakes. e.g. Image recognition, which can be fooled by a pattern on a T-sihrt [slashdot.org]. If your job is completely deterministic
  • Deep blue wasn't very smart. It just could grind through a lot of possibilities. It wasn't even that good at pruning bad moves early it was more of a brute force effort. Even then IBM cheated in that Kasparov was never given examples of games deep blue had played. I suspect with proper preparation Kasparov could have found a style of play that would beat 1997 Deep Blue. He couldn't beat the 2020 chess AIs but they are so much better now and AI has gotten better not just in brute force computation and m

  • Stupid. (Score:3)

    by RightSaidFred99 ( 874576 ) on Friday February 21, 2020 @03:05PM (#59751864)
    This is like asking Usain Bolt feels knowing that a car can beat him in a fucking race. Computers can or will be able to beat a human in any game which can be 'computed'. Durr.

    • Re: (Score:2)

      by MobyDisk ( 75490 )

      Usain Bolt was born into a world where cars already existed. There was no debate as to which was faster: man or machine. He didn't go from being the fastest thing on the planet to being the poster child for why machines are faster. That happened 200 years ago, back when we raced a horse against a train.

    • Re:Stupid. (Score:5, Insightful)

      by MobyDisk ( 75490 ) on Friday February 21, 2020 @04:15PM (#59752128) Homepage

      Sorry, for replying twice, but I just want to add one more thing:

      Computers can or will be able to beat a human in any game which can be 'computed'. Durr.

      You say "Durr" now, but back in 1997, it was common belief was there there were some computation problems that a computer would never be able to handle. And for decades, Chess was seen as the penultimate thinking man's game. The idea that a computer could be a grandmaster was shattering. It was this event that changed everyone's mind. Even 5 years later, people just said "Oh, that's because Chess is really easy for a computer. NOW try that with Poker..." then when they beat us at Poker people said "But Go has so many more branches, computers are terrible at that..." then when they beat us at Go people said "Humans still beat computers at StarCraft..." then it was FPS games... now people here on Slashdot still bleat that self-driving cars will never happen or that it isn't "real" AI.

      While it seems obvious to many people now, the idea that computers could be programmed to solve these things was controversial for quite a long time.

      I can foresee some coot in 2097 saying "That's not *real* AI, of course a computer can run a Fortune 500 company while playing racketball at the same time! But a *real* AI would be able to get my kids to do their Math homework on time! huh huh huh smug chuckle."

      • When computers play StarCraft, they cheat.
        • OK, is this where you want to plant the goalposts, and keep them there? If a robot can beat the best humans at StarCraft by using cameras to look at the monitor and using mechanical hands on a keyboard and mouse, then it will be intelligent?
          • Computers could beat humans in StarCraft a decade ago, it doesn't require any AI. We know this. As to where the goalpost is, if it's not strong AI, then it's not strong AI. Doesn't matter if it can play chess or shoji or paint pictures. If that seems confusing to you, then look up "strong AI" on Wikipedia.

        • Re: (Score:2)

          by MobyDisk ( 75490 )

          Not any more. They put click rate limits on them, and they can only see what fits on the "screen." I thought they had ones now that had to read the screen like a human, but I can't find anything on that so I'm probably wrong there. But gone are the days where they saw the entire map and could do an unlimited number of clicks per second.

    • The fact that back in the 80's Garry Kasparov could play a simul against 32 computers and win all games is no proof of the strength of human brain. It is proof of the pathetic power of computers back in the 80's.
      In fact Garry Kasparov played many games against computers, starting in about 1994 when Intel was the main sponsor of PCA, the federation that he and Nigel Short started when they split from FIDE. And Garry did lose individual games against computers back then, just not entire matches.
      Also, he lost

  • It is as though Usain Bolt were depressed because he stands no chance at a 100m dash against a Tesla car. The fact is that chess, being the finite, simple rules-bound game that it is, is a very similar case. And it does not change things for human chess players in a significant way - just as Bolt does not compete against cars, human chess players will not compete against computers.
    • Actually it changed things for human players in a vast way - because now the specter of cheating hangs over every competition.

      • Oh, bullshit, cheating is actively and effectively countered.

        I'm only rated around 2000, but at all levels computers have changed the way chess is played, changed which openings are considered good and bad, changed how people defend bad positions, changed endgames, changed post-game evaluation, training methods, everything.

        Even timing has been changed by computerized clocks, with most games having additional time added per move instead of the old method of adding time after 40 moves.

  • Yang was probably the best candidate for president we have had in e very long time, beyond his UBI he was talking all the time about this very issue, which will come upon society sooner rather than later.

    Will probably just write him in anyway.

    • Yang was probably the best candidate for president we have had in e very long time, beyond his UBI he was talking all the time about this very issue, which will come upon society sooner rather than later.

      Will probably just write him in anyway.

      Sadly I'm not so sure. I've read articles about free trade where the fine print is that it will be a loss for some but the gains are large enough that even with compensating those who lose it's still a win. What happened - free trade was implemented in such a way that the middle class was gutted and no losers were compensated. Meanwhile inequality soars. Not unlike articles I've read that said that open borders would be great as long as the downsides were managed (ie avoiding swamping votes, keeping maj

Slashdot Top Deals

My sister opened a computer store in Hawaii. She sells C shells down by the seashore.

Close