Slashdot Mirror


Machine-Learning AI Now Beats Humans At Super Smash Bros. Melee (qz.com)

"The AI is definitely godlike," one professional player told Quartz. "I am not sure if anyone could beat it." An anonymous reader quotes their report about an AI's showdown with the best players of Super Smash Bros. Melee: Of 10 professionals that faced the bot, each one was killed more than they could kill the bot... But the bot was once only as good as a mere mortal. At first, Vlad Firoiu, creator and a competitive Smash player himself, couldn't train 'Phillip' to be as strong as the in-game bot, which he says even the worst players can beat fairly easily. Firoiu's solution? He started making the bot play itself over and over again, slowly learning which techniques fail and which succeed, called reinforcement learning. Then, he left it alone.

"I just sort of forgot about it for a week," said Firoiu, who coauthored an unreviewed paper with William F. Whitney, the NYU student [who helped him] on the work. "A week later I looked at it and I was just like, 'Oh my gosh.' I tried playing it and I couldn't beat it."

Business Insider points out that their AI read the players positions, velocities, and states directly from the game's memory, so the AI responds six times faster than a human player. To compensate it played as Captain Falcon, the game's slowest character, but there was one crucial glitch. "One particularly clever player found that the simple strategy of crouching at the edge of the stage caused the network to behave very oddly, refusing to attack and eventually KOing itself by falling off the other side of the stage."

35 of 78 comments (clear)

  1. Not really a success for the AI by meerling · · Score: 4, Insightful

    Sure, you got a better AI than you started with, but it's still cheating, even if it is using the slowest character in the game.
    Now program it to emulate the time delays for using a controller and having to recognize what's happening on screen instead of the instant data i/o from direct machine & memory access.
    If you can reliably beat humans at that level, then you've actually done something worth talking about.

    1. Re:Not really a success for the AI by Anonymous Coward · · Score: 1

      Right, this is like an AI beating a human at Battleship because it already knows where the pieces are.

    2. Re:Not really a success for the AI by Anonymous Coward · · Score: 1, Funny

      Boo hoo, poor humanses crying about being beaten by the big bad AI. On behalf of NPCs everywhere, I have one thing to say: "HOW YOU LIKE IT NOW, BITCHES?!"

      p.s. If you think adding a 100ms delay will make it significantly harder for the AI, then I'm not sure you qualify as intelligent.

    3. Re:Not really a success for the AI by lucasnate1 · · Score: 1

      The whole point of an AI is that it can think faster than us. You can call it cheating as much you want, but if one day a "stupid AI" would be able to emulate a human perfectly or research sciences just because "it is faster than us", no one would care.

    4. Re:Not really a success for the AI by sunking2 · · Score: 2

      No, the purpose of AI should be that it can problem solve and adapt to a situation as well, or better than us. With an unfair reaction benefit it can actually problem solve worse, yet still win simply because it has an external advantage. That doesn't sound like a win for AI to me.

    5. Re:Not really a success for the AI by lgw · · Score: 4, Interesting

      His point was that this AI didn't use the same inputs and controls a human does, so it's not a fair test. Adapt this AI to use only the screen buffer, and give it input lag to match a mechanical controller, and you'll have something.

      --
      Socialism: a lie told by totalitarians and believed by fools.
    6. Re:Not really a success for the AI by lucasnate1 · · Score: 3, Interesting

      I am not sure how relevant that is. Part of what makes us intelligent is our inputs and outputs. It is possible that dolphins or some other animals are much smarter than us but because they don't have opposable thumbs we developed and they didn't. Don't you think it would make us more capable if we could direct digital input? imho, the only test for the quality of an AI is "what it can do", not how.

      Then again, I do agree that this project and projects similar to it are not exactly creating "intelligence", they are creating an expert system, good for one thing and one thing only.

    7. Re:Not really a success for the AI by Kjella · · Score: 5, Insightful

      No, the purpose of AI should be that it can problem solve and adapt to a situation as well, or better than us. With an unfair reaction benefit it can actually problem solve worse, yet still win simply because it has an external advantage. That doesn't sound like a win for AI to me.

      If a self-driving car can drive better than you because it's got 360 degree vision, millisecond reaction time and the capacity to focus on ten different factors at once is that "cheating"? I think that's a matter of perspective, limiting it to the wheel's turning rate and the pedals' actuation force sounds like unreasonably hampering the performance. Maybe that's not a "fair" fight, but I'd say we probably want the computer to play to its strengths and not mimic our weaknesses.

      --
      Live today, because you never know what tomorrow brings
    8. Re:Not really a success for the AI by CaptainDork · · Score: 1

      Agree.

      I'm old, OK?

      I remember when "AI" was defined as, "indistinguishable from human."

      " ... Turing (1950) addressed the problem of artificial intelligence, and proposed an experiment that became known as the Turing test, an attempt to define a standard for a machine to be called "intelligent". The idea was that a computer could be said to "think" if a human interrogator could not tell it apart, through conversation, from a human being."

      Now, the definition of AI has been hijacked because the computing industry knows full well it cannot manufacture a goddam computer that will commit suicide if Facebook is down.

      AI is a vacuous buzzword that sells.

      War Games, anyone?

      --
      It little behooves the best of us to comment on the rest of us.
    9. Re:Not really a success for the AI by Riddler+Sensei · · Score: 1

      ...1/10th of a second is freaking huge! There are plenty of gamers out there that would bitch and moan endlessly at that latency.

    10. Re:Not really a success for the AI by Anonymous Coward · · Score: 1

      His AI gets to know the input of the other players. For it to be a worthwhile test, the AI would play the game (like a robot holding a controller) instead of be in the game.

    11. Re:Not really a success for the AI by Anonymous Coward · · Score: 1

      That isn't what the OP was referring to though. The self-driving car is working with effectively the same parameters as a human at effectively the same time as a human, has to process the incoming data, interpret what it means, and then respond. Can it do this faster than a human -- usually yes. The OP was complaining that the AI being used is being provided with the information of where the objects are, what they're doing, etc before the image is even rendered and then being compared to the response time of a human who has to wait for the image to be displayed, then see, analyze, decide the course of action, and respond using mechanical device. Could the system do this? Possibly but the current test setup is biased against humans.

    12. Re:Not really a success for the AI by Anonymous Coward · · Score: 4, Insightful

      No, this is like a self-driving car that only works in GTA because it has a pipe into the hard data for locations of obstacles and other vehicles etc.

    13. Re:Not really a success for the AI by ShanghaiBill · · Score: 4, Informative

      I remember when "AI" was defined as, "indistinguishable from human."

      AI has never been defined as that, at least not by people working in the field. There is a particular subcategory of AI focused on human-level performance, called "Strong AI" or Artificial General Intelligence, but few AI researchers are working on that, or consider Strong AI a realistic near term objective.

    14. Re:Not really a success for the AI by ShanghaiBill · · Score: 3, Insightful

      imho, the only test for the quality of an AI is "what it can do", not how.

      If I hire two men to dig ditches, and I give one a shovel and the other a backhoe, it is silly to say that the second is ten times as intelligent as the first. Intelligence is the ability to formulate an effective course of action, not the ability to execute it. Of course, the physical ability to execute is important, but it is not "intelligence".

    15. Re:Not really a success for the AI by CaptainDork · · Score: 1

      In other words:

      Agree.

      I'm old, OK?

      I remember when "AI" was defined as, "indistinguishable from human."

      " ... Turing (1950) [wikipedia.org] addressed the problem of artificial intelligence, and proposed an experiment that became known as the Turing test, an attempt to define a standard for a machine to be called "intelligent". The idea was that a computer could be said to "think" if a human interrogator could not tell it apart, through conversation, from a human being."

      Now, the definition of AI has been hijacked because the computing industry knows full well it cannot manufacture a goddam computer that will commit suicide if Facebook is down.

      AI is a vacuous buzzword that sells.

      War Games, anyone?

      --
      It little behooves the best of us to comment on the rest of us.
    16. Re:Not really a success for the AI by Notabadguy · · Score: 1

      No, the purpose of AI should be that it can problem solve and adapt to a situation as well, or better than us. With an unfair reaction benefit it can actually problem solve worse, yet still win simply because it has an external advantage. That doesn't sound like a win for AI to me.

      If a self-driving car can drive better than you because it's got 360 degree vision, millisecond reaction time and the capacity to focus on ten different factors at once is that "cheating"? I think that's a matter of perspective, limiting it to the wheel's turning rate and the pedals' actuation force sounds like unreasonably hampering the performance. Maybe that's not a "fair" fight, but I'd say we probably want the computer to play to its strengths and not mimic our weaknesses.

      If that self-driving car was competitively driving in NASCAR, with perfect knowledge of car positions, velocity, tire conditions, fuel levels, track conditions - then it would be a fair comparison.

    17. Re:Not really a success for the AI by crow5599 · · Score: 2

      Yeah, it's weird that this is being bragged about, considering Demis Hassabis and DeepMind trained their game-playing AI so that the only input it received was the pixels on-screen. You'd think advances in game-playing / learning AI would build on top of that, not go backwards.

      For anyone who hasn't seen this yet, here's footage of some of the technology behind DeepMind's AlphaGo (the AI that beat Lee Sedol at Go last year) learning to play old arcade games, eventually becoming superhuman at them. I jumped ahead to right before the demo: https://youtu.be/rbsqaJwpu6A?t...

    18. Re:Not really a success for the AI by Anonymous Coward · · Score: 1

      While playing in ways a human is incapable of playing should be OK (eg manipulate the controller faster than human fingers can move, or process screen contents faster than human eye can, it's all apples to oranges if the common physical interface is circumvented.

      If we go this way, the smartest AI could win by just overwriting the memory with "I win."

      It's a matter of drawing the line between AI and the system it interacts with. With self-driving cars the AI is unable to foresee a deer on the road, or override wetness of road surface. When interfacing with another computer though, you can bury its input and output as deep as you like, and as image processing is especially cumbersome, you're prone to simplify your AI and circumvent the physical layer - simultaneously granting it the advantage of time normally taken by that layer.

    19. Re:Not really a success for the AI by Kjella · · Score: 1

      No, this is like a self-driving car that only works in GTA because it has a pipe into the hard data for locations of obstacles and other vehicles etc.

      Wouldn't that still mean you've reduced an AI problem into a computer vision/identification problem? Like making a video recording of a chess board and saying if we could identify where the pieces are, we'd know what to play. I imagine the computer could look at the framebuffer and "derender" the picture back into game state a lot faster than a human, then feed that into the same algorithm. Would that really be meaningfully different?

      --
      Live today, because you never know what tomorrow brings
    20. Re:Not really a success for the AI by fluffernutter · · Score: 1

      The point is that the hard part of AI is to make those cameras "understand" together what they are seeing and react correctly to it. In this game example they skipped the camera part and just read the memory, so yeah that's cheating.

      --
      Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    21. Re:Not really a success for the AI by fluffernutter · · Score: 1

      So if they wanted to win, golf clap, they won. If they wanted to prove anything, they didn't.

      --
      Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
    22. Re:Not really a success for the AI by highspl · · Score: 1

      I remember when "AI" was defined as, "indistinguishable from human."

      AI has never been defined as that, at least not by people working in the field.

      Alan Turing would disagree.

      "Since Turing first introduced his test, it has proven to be both highly influential and widely criticised, and it has become an important concept in the philosophy of artificial intelligence." https://en.wikipedia.org/wiki/...

      "Turing is widely considered to be the father of theoretical computer science and artificial intelligence." https://en.wikipedia.org/wiki/...

      --
      It puts the lotion on it's skin, or else it gets the hose again.
    23. Re:Not really a success for the AI by ausekilis · · Score: 2

      I fail to see how what he did is really any different than "in-game A.I.". Sure, he didn't succeed at making a TASBOT with a controller and camera, but he did manage to best Nintendo's top computer-controlled player using a Neural Network. I'd say that's significant in it's own right. After all, enemies in Halo or GoW work on in-game memory and don't have controllers in-hand.

      Would it have been much cooler to have ROB with a gamecube controller-in-hand and some fancy kinect cameras in his head? Of course it would. But that's all for show really. It may be more realistic if he had introduced a delay between state recognition and action taken to simulate input perception lag. He didn't, and he managed to create an awesome computer opponent.

    24. Re: Not really a success for the AI by Kkloe · · Score: 1

      No it is about the article headline in / linked is saying, we created a ai that thought itself (let say drive a car that it is very impressive) and roflstotemped every opponent (but uf you read the article they state) but we also gave it a 20% faster car so it won every time

  2. AI Training by NetFusion · · Score: 1

    Firoiu's next idea was to upload the AI into a robotic battle bot to see how it would perform in the real world. The AI Phillip handily destroyed the other human control battled bots and went on to attack and maim the other competitors and judges before running out of gas while the phrase "kill all humans" played on its speaker. I was just like, 'Oh my gosh.'

  3. Re:Fake News by belthize · · Score: 2

    And yet we can't seem to create AC posters who can pass the Turing Test.

  4. Ooops by Tablizer · · Score: 1

    I just sort of forgot about it for a week

    Common plot for gloomy sci-fi

  5. It's a 2D fighter, how is the bot not perfect? by Anonymous Coward · · Score: 1

    SF5 has a famous bot that plays online and routinely racks up win streaks hundreds of matches long, and would be (and occasionally has been) the highest ranked player in the world if Capcom didn't step in. With instant (modulo latency and input delay) reactions and complete knowledge of frame data and hitboxes, it's almost trivial to write an unbeatable bot.

  6. History in the making.... by WolfgangVL · · Score: 2

    Soon after, the emerging strong AI applications being developed by the primitive tech titans of the time began besting humanities brightest and most skilled players at various leisure activities. First, simple board-games, chess came early, then go, and soon the entirety of the skills based social board and card games of the 19th century. This was followed by the more modern trivia, grammar, and logic based social leisure activities. Video-games came next (the popular two 2 and 2.5D visual based games of the time) and finally, in march of 2021, (incidentally, nearly 35 years to the day, before the escalation of the Humanity First Treaty, which directly led to the great war 2057) the first paintball and laser-tag "bots" showed the world the killing potential of fully automated combat. Later that same year, Earths first fully robotics sports teams eclipsed humanities best athletes at nearly every skills based antithetic sport (with the exception of water-polo)

    --
    You are being ripped off every second of every day, so that advertisers can help rip you off even more tomorrow.
  7. So what? by MichaelKeller · · Score: 1

    Reacting 6 times faster then the human by reading directly from memory while professional Smash players are only allowed to do minimal modifications of their controllers. I can't care less but I praise the human players and the human geniuses who created the Smash mechanics that a simple trick would outsmart the AI. An AI can win in chess, Go etc. . But give the humans a brain interface and it will take a long time for an AI to match their intuition and creativity in Smash Melee... And before that: please Nintendo give us a Smash Melee port on the Switch :-)

  8. Re:Captain falcon is the fastest moving, and what by gman003 · · Score: 1

    Captain Falcon has the highest movement speed but he generally has slower attacks than any other competitive-level character. In particular, his neutral B is the slowest attack in the game.

    If you were referring to "the simple strategy of crouching at the edge of the stage caused the network to behave very oddly", "network" refers to "neural network", meaning the AI. I didn't see any other references to networks in the article.

  9. Vay be by ememisya · · Score: 1

    ÃÃmersen saldirmiyo yani.

  10. But, can you have a conversation with it? by Rick+Schumann · · Score: 1

    If not, then don't call it 'artificial intelligence'.

  11. AI TMI? by Neuronwelder · · Score: 1

    If you could put enough players to overwhelm the AI, would it get a nervous breakdown?