Slashdot Mirror
Games That Design Themselves
destinyland writes "MIT's Media Lab is building 'a game that designs its own AI agents by observing the behavior of humans.' Their ultimate goal? 'Collective AI-driven agents that can interact and converse with humans without requiring programming or specialists to hand-craft behavior and dialogue.' With a similar project underway by a University of California professor, we may soon see radically different games that can 'react with human-like adaptability to whatever situation they're thrust into.'"
I can see it now.
Just before loosing, the AI will suddenly shout "RAGEQUIT" and disconnect, thus denying you points for winning.
switch (last_player_action) {
case QUIT:
exit(0);
default:
move_pitiful_player_char(last_player_action.direction, LUDICROUS_SPEED);
ai.queue.append(last_player_action);
ai.queue.append(new_action(ACTION_SAY_TO, player, "quit following me!"));
}
Artificial intelligence came a step closer this weekend when an MIT computer game, which learnt from imitating humans on the Internet, came within five percent of passing the Turing Test, which the computer passes if people cannot tell between the computer and a human.
The winning conversation was with competitor LOLBOT:
The human tester said he couldn't believe a computer could be so mind-numbingly stupid.
LOLBOT has since been released into the wild to post random abuse, hentai manga and titty shots to 4chan, after having been banned from YouTube for commenting in a perspicacious and on-topic manner.
LOLBOT was also preemptively banned from editing Wikipedia. "We don't consider this sort of thing a suitable use of the encyclopedia," sniffed administrator WikiFiddler451, who said it had nothing to do with his having been one of the human test subjects picked as a computer.
"This is a marvellous achievement, and shows great progress toward goals I've worked for all my life," said Professor Kevin Warwick of the University of Reading, confirming his status as a system failing the Turing test.
http://rocknerd.co.uk
The idea of an AI that learns from the players sounds great when you're talking about a bot for Multiplayer Shooter 2010 developing tactics and strategies without explicit programming, or an NPC partner in a stealth gaming learning how not to bash their face into walls and then walk off a cliff into lava. Awesome, bring on the learned emergent behavior!
But dialogue? Oh lord no, please don't let the AI's learn how to "converse" from players. Because the last thing I need is to have AIs in games screaming "Shitcock!" or calling me a fag a thousand times in a row with computerized speed and efficiency.
The enemies of Democracy are
This one shouldn't be too hard.
Learning HOW to think is more important than learning WHAT to think.
Ask WoW developer, they can't spot most of the bots playing the game.
Love many, trust a few, do harm to none.
Well, if it learns from me there's gonna be a screen full of Agent Smith's beating off to a screen full of Jessica Alba's.
http://xkcd.com/117/
Does this mean somebody's porting Eliza to Ruby on Rails?!
I always thought it would be interesting to create a project like this with a chat engine. Take a major chat engine and have a "Submit to AI" option where the AI would parse the conversation between you and a friend so it can record questions and responses in an overlapping matrix of possibilities and calculate the probability of what the response should be by historical conversations of the same nature. You should get impressive test results with a large enough set of data.
CS: It is all sink or swim...oh and did I mention there are sharks in that water?
Because programming -IS- Logic. If you tell the program to do soemthing at Random, its not a very good AI. If you tell it to do the most strategically sound plan, it doesn't vary much at all.
You tell it to try to learn the rules, and make the best decision that it can.
Consider AI for chess. The best AI can beat any human because it can spend the processing power to look, say, 25 moves into the future. When the computer considers all possible moves and for each one looks at all possible next moves, next moves, etc, for 25 turns, it's going to be able to quantify which move it should make now to have the best chance at winning. When you download a chess game and you can set the difficulty, the main thing they change is how far ahead the AI is allowed to look. An "easy" AI might only look 3 moves ahead. It's been a while since I took any AI courses, but I seem to remember that the human masters like Kasparov are capable of looking ahead around 10-12 turns.
So it's not that you tell the AI to make bad decisions, you simply limit the information it has to work with. This is more equivalent to what most humans do when they make bad decisions ("I didn't think of that").
"Our two-party system is like a bowl of shit looking at itself in a mirror." - Lewis Black
Because programming -IS- Logic.
A group of neurons can be connected together to form a calculator. But, you can't multiply 20 digit numbers in your head. You don't have access to the "hardware" layer of your brain. Why would a sufficiently advanced AI be any different?
As such you generally tend to base it against the opponent you are playing. An AI cannot tell if you are an aggressive or passive person, you're strategic abilities or understanding of game mechanics having never met you before playing the game.
I play online games against people I've never met before too. What magical ability do I have, that a computer could not?
... and that's when the C.H.U.D.'s came at me.
I we humans
Except when it comes to using the English language, apparently.
Everything Peter does looks impressive while he stands by it. He's like a lesser powered Steve Jobs. However, unlike Steve, Peter's glamour effect only lasts till the product is released. Should Milo ever actually hit the market, it will immediately revert to a simulation of an autistic Eliza with Turrets syndrome and a tendency to stare at crotch rather than your face.
Peter will then appear and indicate that he knew Milo I was going to be this bad, that's why for the past TWO decades, he's been working on Milo II, which will suppose to do everything he actually promised in Milo I and include a loveable dog character for you to interact with as well.
When Milo II finally comes out, it'll be an actual stuffed basset hound.
Shit, when I played WoW I spent lots of time trying to get a /follow train to completely encircle Ironforge.
Never got a full train (a circle of people following each other, where the "engine" eventually is close to the "caboose" and does a /follow on them) though...
Indeed.
In fact, feeding bogus data to the AI is one of the realistic ways to limit, say, a racing game's agents - if they don't see the post in front of them because they aren't spending enough time per frame watching the road and are instead eyeballing their opponent, they're going to crash, just like any human. So you simulate that by using player proximity and the "erraticness" of the other opponents to model distraction and modulate the AI's awareness of dynamic obstacles and hazards.
The land shall stone them with the bread of his son.
I'm not sure what's worse... that you could write that without collapsing, or that I could actually hear it in a perfect valley girl voice.
Yes and no. Back in the day when I was writing Quake bots, there were things you could do to always beat the AI. The AI cant pick out patterns that are luring it into a trap. WE are a long LONG way from having AI that can think about the situation and make a decision on it's own...
"Player 4 has done this 4 times trying to lead me down that corridor, what the hell is he doing? I'm gonna sit and wait or try and circle around to see what is up."
AI cant make a conscious decision that is not preprogrammed.
Do not look at laser with remaining good eye.
A recent proposal from the UC Santa Cruz EIS lab (also mentioned in the article), an Automated Game Designer: http://www.slideshare.net/rndmcnlly/the-intelligent-game-design-game-design-as-a-new-domain-for-automated-discovery-1784151 It's not about making a bot that can behave intelligently/interestingly in a restaurant setting... what are the broad applications of that? (As other people have pointed out, the bots may come out pretty demented and flavored like The Internet.) It's about making a game designer that can design games on its own, learn from its own experience, get MINIMAL human input (not 10,000 plays online). The computer designer can do what the computer is good at (enumerate all possible play traces, look for instances of accessibility/cheats/funky behavior the designer might not have intended or expected) and the humans on the side can do what they are good at (shaping, polishing, collecting a few human play traces).
So they're stealing our body heat and letting us write agent AI for them too? Geez, what lazy AI we invented.
It was created in our own image.
Track your TV Shows with your iPhone - FREE
Are there any examples of a living being which does not spend the majority of it's life parroting or applying the behaviour of others?
I'd contend that watching and mimicing others is the most effective method of learning. In fact, it's the ability to take and apply this learned knowledge to other situations that seperates the truly intelligent from the "average" in the world.
Just another ignorant American.
The problem with AI's mimicking 'human' actions has nothing to do with a failure of logic or the ability to display randomness.
It has to do with the fact that we've never really understood why we do certain things, because we hold the false notion that for the most part our actions are driven by logic rather than the reality that our logic is driven by our actions. Thus, when someone happens that doesn't fit our model, we ascribe it to randomness despite the fact that it could probably be shown that the same situation would result in the same reaction the majority of the time.
If we actually studied our actions, rather than our rationalizations for why they occur, it'd be a lot easier to model our behavior.
And an AI that doesn't care why we do something but just learns to predict WHAT we will do, is a good first step towards that.
If the AI Agents are learning to mimic human behavior by observing how they play a game, then the game design clearly already exists. Therefore, what is described in the article is certainly not anything even remotely close to "games that design themselves."
AI cant make a conscious decision that is not preprogrammed.
That's not true. Look at the PROLOG language, or LISP. You don't need to program all possible decisions into an agent, you just need to give it the capacity to learn and assign various weights and things to the things it thinks are important so that it can quantify what the best decision is. With PROLOG specifically you can give an agent the ability to draw new conclusions based on things it already knows (which it then adds to its list of things that it knows).
We're not as far from this as you might think..
"Our two-party system is like a bowl of shit looking at itself in a mirror." - Lewis Black
I'd say something snarky, but that would require effort.
I don't suffer from insanity, I enjoy every minute of it!
And a computer could not have consciousness because...
... and that's when the C.H.U.D.'s came at me.
man hasn't "(re-)invented" it yet, and isn't likely to for a long time to come.
That "long time" will be forever, if we never research it. You've got to start somewhere.
Do you think we'll just magically come up with the answer, if we never think about the question?
... and that's when the C.H.U.D.'s came at me.