Opponents can be done reasonably well with even the simple AI we have now. The killer app for gaming would be AI characters who can respond meaningfully to the player talking to them, at the level of actually generating new prewritten game plot quality responses based on the stuff the player comes up with during the game.
This is quite different from chatbots and their ilk, I’m thinking of complex, multiagent player-instigated plots such as the player convincing AI NPC A to disguise itself as AI NPC B to fool AI NPC C who is expecting to interact with B, all without the game developer having anticipated that this can be done and without the player feeling like they have gone from playing a story game to hacking AI code.
So I do see a case here. The game industry has thus far been very conservative about weird AI techniques, but since cutting edge visuals seem to be approaching diminishing returns, there could be room for a gamedev enterprise going for something very different. The big problem is that when sorta-there visuals can be pretty impressive, sorta there general NPC AI will probably look quite weird and stupid in a game plot.
Opponents can be done reasonably well with even the simple AI we have now.
Not for games like Civilization they can’t. Especially not if they’re also supposed to deal with mods that add entirely new features.
Some EURISKO-type engine that could play a lot of games against itself and then come up with good strategies (and which could be rerun after each rules change) would be a huge step forward.
It would be very bad if an opponent AI went FOOM. Or even one which optimized for certain types of “fun”, say, rescue scenarios.
But consider a game AI which optimized for features found in some games today (generalized):
The challenges of many games require you to learn to think faster as the game progresses.
They often require you to know more (and learn to transfer that knowledge, part of what I would call “thinking better”).
Through roleplaying and story, some games lead you to act the part of a person more like who you wish you were.
Many social games encourage you to rapidly develop skills in cooperation and teamwork, to exchange trust and empathy in and out of the game. They want you to catch up to the players who already have an advantage: those who had grown up farther together.
There are more conditions to CEV as usually stated, and they are hard to correlate with goals that any existing game designers consciously implement. They might have to be a hard pitch, “social innovations” for a “revolutionary game”.
If it was done consciously, it’s conceivable that AI researchers could use game funding to implement Friendly AGI.
(Has there been a post or discussion yet on designing a Game AI that implements CEV? If so, I must read it. If not, I will write it.)
Needless to say having one of these go FOOM would be very, very bad.
Maybe, but the purpose of such an opponent isn’t to crush humans, it’s to give them as good a game as possible. The big risk might be an AI which is inveigling people into playing the game more than is good for them, leading to a world which is indistinguishable from a world in which humans are competing to invent better superstimulus games.
AI opponents seem to have a relatively easy timing defeating human players at many games already.
I think it’s possible that the development of AI players that are more fun to play with or against will be a new direction for gaming AI to go which would be far less tragic (compared with, say, Astonishing X-men volume 3 issues 8-15+)
Of course this is just a possibility, I don’t mean to say that this is the most likely outcome.
I think that’s been happening for a while already. I vaguely remember reading somewhere that the main difficulty in designing AI game opponents was making them stupid enough to beat.
I assume you mean designing better AI opponents, as this seems to be one type of very convenient problem for AI.
Needless to say having one of these go FOOM would be very, very bad.
Opponents can be done reasonably well with even the simple AI we have now. The killer app for gaming would be AI characters who can respond meaningfully to the player talking to them, at the level of actually generating new prewritten game plot quality responses based on the stuff the player comes up with during the game.
This is quite different from chatbots and their ilk, I’m thinking of complex, multiagent player-instigated plots such as the player convincing AI NPC A to disguise itself as AI NPC B to fool AI NPC C who is expecting to interact with B, all without the game developer having anticipated that this can be done and without the player feeling like they have gone from playing a story game to hacking AI code.
So I do see a case here. The game industry has thus far been very conservative about weird AI techniques, but since cutting edge visuals seem to be approaching diminishing returns, there could be room for a gamedev enterprise going for something very different. The big problem is that when sorta-there visuals can be pretty impressive, sorta there general NPC AI will probably look quite weird and stupid in a game plot.
Not for games like Civilization they can’t. Especially not if they’re also supposed to deal with mods that add entirely new features.
Some EURISKO-type engine that could play a lot of games against itself and then come up with good strategies (and which could be rerun after each rules change) would be a huge step forward.
This is what I was trying to say but much better.
It would be very bad if an opponent AI went FOOM. Or even one which optimized for certain types of “fun”, say, rescue scenarios.
But consider a game AI which optimized for features found in some games today (generalized):
The challenges of many games require you to learn to think faster as the game progresses.
They often require you to know more (and learn to transfer that knowledge, part of what I would call “thinking better”).
Through roleplaying and story, some games lead you to act the part of a person more like who you wish you were.
Many social games encourage you to rapidly develop skills in cooperation and teamwork, to exchange trust and empathy in and out of the game. They want you to catch up to the players who already have an advantage: those who had grown up farther together.
There are more conditions to CEV as usually stated, and they are hard to correlate with goals that any existing game designers consciously implement. They might have to be a hard pitch, “social innovations” for a “revolutionary game”.
If it was done consciously, it’s conceivable that AI researchers could use game funding to implement Friendly AGI.
(Has there been a post or discussion yet on designing a Game AI that implements CEV? If so, I must read it. If not, I will write it.)
Needless to say having one of these go FOOM would be very, very bad.
Maybe, but the purpose of such an opponent isn’t to crush humans, it’s to give them as good a game as possible. The big risk might be an AI which is inveigling people into playing the game more than is good for them, leading to a world which is indistinguishable from a world in which humans are competing to invent better superstimulus games.
eh, given the space of various possible futures I would regard this as one of the better ones.
AI opponents seem to have a relatively easy timing defeating human players at many games already.
I think it’s possible that the development of AI players that are more fun to play with or against will be a new direction for gaming AI to go which would be far less tragic (compared with, say, Astonishing X-men volume 3 issues 8-15+)
Of course this is just a possibility, I don’t mean to say that this is the most likely outcome.
I think that’s been happening for a while already. I vaguely remember reading somewhere that the main difficulty in designing AI game opponents was making them stupid enough to beat.