Slashdot Mirror
Game AI Conference Explored
Academia Blog Grand Text Auto has up a long set of notes from last week's first AI and Interactive Entertainment conference, which includes keynote talks from Doug Church, Will Wright, Chris Crawford and Damian Isla of Halo 2. From the Doug Church talk: "none of the AI detail gets attention in a 30 second ad or magazine blurb...also, if a character in battle only lives a minute, there's not much fidelity players can even perceive...industry has been promising good characters for a long time, not delivered...
players are cynical, don't want to hear it anymore...hard to back out of the fakery"
The quote is kind of hard to understand. I RTFA and the point of that quote still eludes me.
If you like what I've said here, and want to read more, go to http://www.krillrblog.com
I wonder if the experiments in simulating the human brain will come in handy with making an AI. I mean if we can understand how the human mind makes decisions, we can better code an AI that can make decisions the way we can based on environment, situation... how we feel about the world... etc.
If you like what I've said here, and want to read more, go to http://www.krillrblog.com
Could they maybe focus on fixing their bugs (slashdot.org) before making a "kick-ass" AI?
If you like what I've said here, and want to read more, go to http://www.krillrblog.com
A link (slashdot.org) to the slashdot article about making a "virtual brain".
I'm starting to feel very silly, because i'm the only one commenting on this story. 4 comments so far too.
If you like what I've said here, and want to read more, go to http://www.krillrblog.com
I'm covering the overall architecture of the brain right now in my intro. psych. class.
It looks remarkably like a computer architecture when broken into components. You have your I/O neurons, interfacing with a component that discards noise. From here it is put onto a bus, into what the textbook labels as the "executive" (CPU). The executive can store and load from short-term memory (registers)---there are different kinds for various senses and parts of cognition---and do other brain-type things. It can also store and retrieve from long-term memory.
It is a computer---just a very, very analog kind.
Actually, if you think about, it may be closer to being a quantum computer than anything else. There aren't 1s and 0s in our brains, but multitudes of different states for each neuron, as there are multitudes of different spin states for each particle in a quantum computer.
If you like what I've said here, and want to read more, go to http://www.krillrblog.com
What? Cause Halo 2 had such good A.I. he has the high ground to insult other peoples' A.I.? Come ON!
It looks a whole lot like a computer because that's the dominant metaphor that we're using and we're filtering our explanations through that idea.
At the time of Descartes, they thought that everything looked kind of like it was clockwork and so they made explanation using that metaphor. Freud did a pretty good job of explaining the psyche in terms of conflicting forces and Jung did it by populating our heads with stereotypes.
Metaphors are cool that way - they are flexible and once you get a halfway good metaphor, you start noticing all kinds of similarities (for example have you noticed that EVERYONE either looks like a pig or a rat? try it!). But it's good to remember that the map is not the territory.
I have a lot of opinions about Cyborgs and Architects
This doesn't make too much sense. The fact that a neuron is more complicated than a bit and can take up component states means nothing. Component states can simply be thought of as single states (something in states A and B is in state A-B), and multiple bits could map to one neuron. There's nothing that says a computer needs to work on bits. A quantum computer is a different beast all together :)
I think AI programmers would be better off studying philosophy, psychology and sociology rather than the mechanics of the brain ;)
"A truly wise man realizes he knows nothing."
It could also be that man's creative impulses have drove us to try to create new lifeforms and the computer is our best attempt yet ;)
"A truly wise man realizes he knows nothing."
An AI that reacted like a human brain in a Half-Life game would wet it's pants. You want an AI that will panic under fire?
Yes.
If you like what I've said here, and want to read more, go to http://www.krillrblog.com
They simply are thinking to small. Smarter enemies, smarter units in RTS games, smarter comrades in team-based games.. Sure - all these things are the realm of AI.. But they are simply thinking too small!
One area of AI-usage that I have not seen explored yet is crafting an AI that would respond to the players actions and modify the flow of the game. In example: suppose I was developing [yet another] fantasy-based MMORPG. Perhaps this one involves 3 kingdoms caught in a never ending struggle (remind you of DAoC perhaps?).. Suppose there was an intelligent AI that created the quests?
Perhaps I get a quest to scout an enemy position. I do this and am spotted by a guard, and a quest is generated by the enemy side to send a group to assassinate me before I can report back. I get back safely and a quest is generated for a group to escort me to the nearest city to ask for reinforcements, all the while the previous group's quest (the ones to assassinate me) is altered to now prevent us from reaching that city. Etc etc
Obviously this doesn't need to be on the MMO-scale to work. If an intelligent AI could be crafted to make your goals and missions accurately reflect your actions in the game then you could theoretically have a game that is totally open ended while still feeling story-driven. A lofty goal, but one I would love to see done. AI of this type could really revolutionize things.
--- "End Of Line" - MCP
Throw in the whole 'well what if there were no other players to escort the original scout player to the city?' issue or the ever-present 'what about balancing the two sides? can't have one side have 10 level 20 guys and the other having 5 level 5 guys' and you've got an unbalanced, arguably buggy MMO game.
A better, simpler example would be if the NPCs gave chase to you DEPENDING on where, when, how and who/which NPC saw you. If you got spotted by NPC Rookie level 1, in the dark, late at night and only saw you for 2 seconds the AI might not even give chase to you. On the other hand if NPC Commander level 50 saw you, in broad daylight, during an important tactical mission for about 15 minutes but didn't say anything to avoid 'spooking' you, the AI SHOULD go berserk sending troops at you with enough man and firepower to rival some player guilds/clans.
industry has been promising good characters for a long time, not delivered
Feature-length films with millions of dollars behind them, professional actors who dedicate months of their lives to portraying a single role, and writers who spend years on a screenplay often fail to deliver good characters. Who in their right minds expects good characters from some lousy computer program?
I wonder if the experiments in simulating the human brain will come in handy with making an AI.
AI for a computer game? Hardly.
One of the most effective, fun game AIs I've played against recently is that in Halo - it's probably no more advanced than that in some other games, but it has some great application of smoke-and-mirrors and does a good job of presenting obvious cause-and-effect behaviours to the player.
Kill a Covenant Elite, and all the lowly grunts nearby will panic and try to run and hide. But to actually shoot that Elite, it's probably taking cover behind a rock, waiting for you to attack - it doesn't just meander into combat, shooting blindly.
All sorts of things like this - simple 'IF foo THEN bar' behaviours which the player can learn, understand and anticipate can be great, so long as they're fun to play against. Some hyper-intelligent enemy that can figure out precisely where the player is and attack unseen might be programmatically more advanced, but isn't necessarily more fun to play against.
In-the-field tactics are probably best left to the game AI, but higher-level, map-specific scripted strategies can give the illusion of some overall plan behind the enemies' actions (plus they can be designed to be fun to fight against, rather than being whatever the AI might extrude - fun, crap or otherwise).
Neural networks or whatever might be more 'realistic', but they won't necessarily be better to play against...
Tedious Bloggy Stuff - hooray?
I'm nor convinced, people even want a strong AI.
I for one like shooting hordes of relatively stupid enemies. I love Resident Evil 4 (finished it yesterday), but the enemies are not exactly clever (and no, that's not because they're zombies. They're not).
The challenge comes mostly from their number and some clever locations they're sitting or waiting.
IMHO, a clever combination of scripts and behaviour patterns should be enough to provide a reasonable challenge.
If people want good (A)I, they can play online.
I don't need a signature.
Although there are a few bright gems in the future that this may be true about, I feel that this statement is for the most part false.
Just look at the PS3 and Xbox 360. Sony and Microsoft have gotten into a dick measuring contest about who can have the best specs. Nintendo is promising a revolution, but I'm guessing it will be more of the same old. Next generation games will be a lot like today's games, only prettier.
Companies aren't going to use the power of next generation consoles to create better content. Square-Enix is just going to animate the individual 50,000 strands of hair on the heads of each character, and Team Ninja is just going to add even more detail and jiggle if you know what I mean.
Eventually things will shift back to games with content, but we'll eventually come full circle and repeat this nonsense all over again. If we've learned anything from our history, it's that we've never learned anything from our history.
Fun, one month ago a friend presented his thesis to the jury.
Thesis is a learning system for game NPCs. He used Team Fortress Capture the flag, one specific map, and trained his 4 bots against FoxBot and another bot which name i don't remember.
After some training, he gathered 4 human players, and made them play against the 3 bot teams (not saying which was which bot, of course !). Results? Human players found his bots were the most realistic and surprising to play with (not the hardest).
Hopefully he'll eventually work in AI for games (that's what his thesis is about, after all !), and make great things.
Tsuyoikoto ha taisetsu da ne, dakedo namida mo hitsuyousa (Strength is an important thing, but tears too are necessary)
### Neural networks or whatever might be more 'realistic', but they won't necessarily be better to play against...
ACK, I see the biggest use of neural networks and advanced AI not so much in enemy behaviour, but in interactive story telling. Today most games run on some basic AI combined with a bunch of prescripted events, so basically everything the player is going to see is already set right from the start. What I think will play a important role in the future aren't more clever enemies, but an a form of semi-automatic story-telling, which doesn't mean that the enemies get clever, but that the fight has a whole will get more interesting and more cinematic. Simple example, if you shoot at an airplane in a game it will explode and crash somewhere, its ok, but not that interesting. In contrast to that what happens in a movie? The hero shoots at some airplane and instead of just crashing somewhere its coming towards the hero and crashing right infront of him, causing him to try to escape or it crashes in some building nearby and causing a bunch of nice explosions, probally making the building falling towards the hero. Games so far don't offer such special events or if they do, they are most likly completly prescripted or pure luck. In future games I imagine some short of 'game master'-AI, much the same of what a game master does when playing a pen&paper RPG, will take care of the overall happenings on the battlefield and make sure that the player gets something interesting to fight, instead of just something more clever and realistic. After all we play games to have fun and real battle isn't fun, so improving the enemy AI alone won't help to create better games.
In addition to yes (as the other poster already mentioned), the HL AI does panic under fire. They don't do it all the time - usually they hold their position and try to you out. If you throw a grenade, then they will try to flee.
The NYTimes wrote an enthusiastic article about the conference.
-Eric
SJW: Someone who has run out of real oppression, and has to fake it.