A.I. – no wait, just I.

I keep reading all these articles in various print and online game magazines where writers lament the continued lack of “realistic” AI in current and future game titles. It seems that next-gen consoles favor graphics pipeline over architecture that might help to improve AI. So the bitch-fest will continue. Regardless of hardware setup on this or the next cycle, players will complain because hardware won’t be able to do what human opponents do so well: go outside the box, do intelligently unexpected things, and think semi-strategically. Right?

Ok, so… what if EA (for example) started putting “EA AI!” stickers on some of their games. Tell players that in order for the advanced AI to work it has to connect with some master server to “download new strategies” or whatever (you get normal AI otherwise)… And guess what? This new AI *really* kicks major ass. It acts like a human opponent, does unexpected things, even seems to taunt and convey emotion in its actions. Awesome! People talk about it. Want more of it. EA licenses it to other publishers, whatever…

And the man behind the curtain: 1000s of kids in China sitting at “dungeon master” consoles controlling the game environment, AI states, and other game variables. Once the news gets out of how it works people will probably want even more of it! If nothing else it creates a serious amount of media coverage. “Star” dungeon masters get created. People pay a premium to “go up against the best”, etc, etc.

Yeah, all this has to be built into the games beforehand, but I’d imagine that these hooks would be a lot easier (and cheaper) to implement than actual AI code to match the complexity and variability from the human “AI machines”. We all know that “kids in China” are happy to sit at computers day & night to mine MMOs for money. Why not have them do something useful: provide AI horsepower for our single player games?!

14 responses to “A.I. – no wait, just I.”

  1. But isn’t that just a half-a-step past the online universes that are already created like EverCrack and UO? How would a “virtual-virtual AI” be different, except that certain players would get paid?

  2. O__O


    Yeah, why not?!

    It’s a verry good idea!

    For the serious thoughs : the part about “star” Ennemi player makes me think that a game with normal players and special players, that don’t have the same rules than normal players, whould be verry difficult to balance, but should provide “stars” in the Game Bosses players…

    Well, thank’s for the reflexion hahaha XD

  3. You know what they say…

    Artificial Intelligence is no match for natural stupidity.

  4. A.I. in games isn’t even that bad. The I in real life is often much worse.

    On the “kids in china” part, excelent fuel for will Wright’s Spore?

  5. When I was really young I used to sometimes wonder if a real person somewhere far off was controlling the enemies in my video games. I swear I’m not that stupid anymore.

  6. Sounds like youre being kept busy!
    I know this is real cheeky and I’m sorry for that but I was wondering how much it would cost if I asked you to design a tattoo for me? and also if you would do it?
    I really adore your work and I totally understand if the answer is no. Can’t blame a girl for trying ! I also model for Suicide Girls and plan to submit a photoset based on your Alice, just incase you would be interested to know.

  7. The problem isn’t even the effectiveness of the AI, its that the models of gameplay are stuck in the old cherodes of spatial action. One of my goals is to do an AI based platform for interactive drama and whatever complimentary sort of gameplay you’d want to design, a platform capable of taking genetic content and recombining it dynamically. Since challenges would be staged on this AI platform, its concievable that it could also balance constraints to be just as challenging as appropriate. Though I think we should stop thinking of challenge as something the player must overcome, and more of a “reality” we want to represent by having the player negotiate it.

  8. Smacks of “A Young Lady’s Illustrated Primer” by Stephenson. Nice to find your blog by the way, I’ll be reading 🙂

  9. Kind of a scary thought. But would it be cheaper to pay an AI programmer $100,000/yr for a year or so to complete some new AI, or pay 1000s of Chinese kids for the rest of their life to be the AI in a game? Ah, I guess it wouldn’t matter much, EA would still be making a boatload of money.

  10. Darn, can’t edit my post. I had another idea. With the general acceptance of graphics cards, sound cards and the rising of physics cards, could AI be left alone to the CPU? I know there are more things the CPU would be doing during the game, but maybe these advances in the other areas could lead to some real innovation in the area of AI. Or, we could just have an AI card…

  11. Actually, when I was about wee-high, I always day-dreamed about something like that, itd be hella tight to be fighting a horde of zombies controlled by some dude somewhere else.

  12. OK, two things:

    First: On the “kids in China” thing, in all seriousness, when I was working with online wallet technology in the late 1990s, we almost partnered with a now-defunct search engine company. While giving us a tour of their offices, they took us into a massive open space, crowded wall-to-wall with Indian programmers. They were taking search engine requests from users, doing searches across other search engines and forums, and spitting results back to the requesting user. No joke. They’re out of business now …

    Second thing: I’m curious how the “ground-breaking AI technology for use in next-generation massively multiplayer online games” from places like Online Alchemy (http://www.onlinealchemy.com/) will pan out. I’ve also wondered that if there are “Physics Engines” (a la Havok) and the like, why there aren’t “AI engines”? Seems like it would be pretty useful, say, if Will Wright or Peter Molyneux could abstract, package, and license “Spore” or “Black and White” AI for repurposing in other titles. Nice little revenue stream, too.

  13. A challenge is always good and all, but no one wants to play against Kasparov-level AI. Challenge is only needed to keep a player motivated. Video games are more about fun and social interaction if you’re playing online games, so it does indeed get a bit annoying to hear all these guys whine and moan about a game not being hard enough. Bad AI is one thing, but if you’ve mastered the gameplay and can easily beat the game on the highest difficulty with little effort, guess what? BUY ANOTHER GAME!
    No point in complaining about a game that you’ve voluntarily exhausted.

Leave a Reply