quote:
Original post by ahw
We are simply making the stakes higher and higher... an arms race, if you will.
quote:
Original post by ferretman
I base this on two reasons. First, developers don''t seem terribly motivated to "beat" each other''s AI, at least not yet. I''ve heard and read lots of articles in which so-and-so will offer "more scenarios", or "more realistic graphics", or whatnot--but I haven''t heard anybody (yet) say "we''re going to beat that game''s AI!".
Second, I''m still having a hard time believing that AI per se in a game like Go or C&C really can compare in terms of intricacy to any decent human player. It''s one thing to beat the human; it''s another to seem human yourself.
Ah, as usual I have problems putting my deep thoughts in words
![](wink.gif)
I meant an arms race between the human and the machine.
Think Deep Blue, here. We have finally beaten Kasparov... yay... big deal ! Does Deep Blue have an interesting conversation? does it make one of those so human mistakes ? Can you play on its nerves to actually win (I know I do that a lot to other players) ?
We simply try to program better and better, offering an increasingly difficult challenge. But just like rats adapt to increasingly lethal poisons, our little brains adapt to the difficulties offered, and tempered anyway by game designers...
quote:
Original post by ferretman
Interesting.....I agree with the need to add more meaning and think it''s interesting you believe this will come from the RTS realm. I would have pegged multiplayer RPGs as a more suitable field for those seeds.
Well, I dont consider that there will anything really big in AI that will come from MMORPG. I am much more convinced that there will be better techniques to deal with data, especially level storage. I am still wondering how the heck they deal with thousands of players on so vast levels... But anyway, the big deal in MMORPG is the environment and other players, so I dont think the AI is all that important. Some human controlled NPC would do. IMHO (but that''s more of a design topic, I talked about this more in GAme Design forum)
Then there are CRPGs. Well, you *would* think that there would be something coming from there. But the plain fact is that I am just not convinced at all. It''s much easier to make a massively scripted game, than a freeform open one. If you have ever played a pen&paper RPG, you''ll know I am right
![](tongue.gif)
Then the choice is between FPS and RTS, right ?
I would love to think that the AI in FPS will be greatly improved, and yada yada yada... but I just don''t see it happen as fast as what it can in RTS.
Those foxy assassins might whoop my ass with their speed, and agility, but they still can''t plan my reactions, and just dont seem to understand that shooting *through* your comrades, even if it''s to reach me, the player, is a bad thing...
Besides, RTS have a simpler environment to deal with. As I said, I don''t think you can improve the AI if you dont start by improving the environment they evolve in. And having a kick ass AI in a 3D game is much harder than in a 2D based game. Not because of the one dimension missing, but because 2d based games tend to have more meaning bound to them. Just think for instance, of the nature of the ground. In an RTS, each cell is attributed some nature that the AI can work upon, knowing they are gonna be slowed down, etc. In a 3D game, because we need to speed up all the rendering, all this ground is a bunch of polygons amongst thousands of others. They do have a texture, maybe some special effect bound to them. But they have no meaning per se. When a 3D AI will be able to look at a bunch of polygons and call them "a door", or "the ground", then I''ll start changing my mind.
And don''t think I am an addict of RTS, I actually hate the damn f*&kers because they are just sooooooo far from playing with miniatures and a bunch of friends on your living room carpet.
But for the moment, I put my bets on RTS
![](smile.gif)
quote:
Original post by ahw
We are simply making the stakes higher and higher... an arms race, if you will.
quote:
Original post by ferretman
I base this on two reasons. First, developers don''t seem terribly motivated to "beat" each other''s AI, at least not yet. I''ve heard and read lots of articles in which so-and-so will offer "more scenarios", or "more realistic graphics", or whatnot--but I haven''t heard anybody (yet) say "we''re going to beat that game''s AI!".
Second, I''m still having a hard time believing that AI per se in a game like Go or C&C really can compare in terms of intricacy to any decent human player. It''s one thing to beat the human; it''s another to seem human yourself.
Ah, as usual I have problems putting my deep thoughts in words
![](wink.gif)
I meant an arms race between the human and the machine.
Think Deep Blue, here. We have finally beaten Kasparov... yay... big deal ! Does Deep Blue have an interesting conversation? does it make one of those so human mistakes ? Can you play on its nerves to actually win (I know I do that a lot to other players) ?
We simply try to program better and better, offering an increasingly difficult challenge. But just like rats adapt to increasingly lethal poisons, our little brains adapt to the difficulties offered, and tempered anyway by game designers...
quote:
Original post by ferretman
Interesting.....I agree with the need to add more meaning and think it''s interesting you believe this will come from the RTS realm. I would have pegged multiplayer RPGs as a more suitable field for those seeds.
Well, I dont consider that there will anything really big in AI that will come from MMORPG. I am much more convinced that there will be better techniques to deal with data, especially level storage. I am still wondering how the heck they deal with thousands of players on so vast levels... But anyway, the big deal in MMORPG is the environment and other players, so I dont think the AI is all that important. Some human controlled NPC would do. IMHO (but that''s more of a design topic, I talked about this more in GAme Design forum)
Then there are CRPGs. Well, you *would* think that there would be something coming from there. But the plain fact is that I am just not convinced at all. It''s much easier to make a massively scripted game, than a freeform open one. If you have ever played a pen&paper RPG, you''ll know I am right
![](tongue.gif)
Then the choice is between FPS and RTS, right ?
I would love to think that the AI in FPS will be greatly improved, and yada yada yada... but I just don''t see it happen as fast as what it can in RTS.
Those foxy assassins might whoop my ass with their speed, and agility, but they still can''t plan my reactions, and just dont seem to understand that shooting *through* your comrades, even if it''s to reach me, the player, is a bad thing...
Besides, RTS have a simpler environment to deal with. As I said, I don''t think you can improve the AI if you dont start by improving the environment they evolve in. And having a kick ass AI in a 3D game is much harder than in a 2D based game. Not because of the one dimension missing, but because 2d based games tend to have more meaning bound to them. Just think for instance, of the nature of the ground. In an RTS, each cell is attributed some nature that the AI can work upon, knowing they are gonna be slowed down, etc. In a 3D game, because we need to speed up all the rendering, all this ground is a bunch of polygons amongst thousands of others. They do have a texture, maybe some special effect bound to them. But they have no meaning per se. When a 3D AI
quote:
Original post by ferretman
It''s a lot of work getting a Master''s...hang in there!
Everytime one of my students comes back to me to thank me for my patience, efforts, and to tell me that "now, I understand", it makes all the difficulties seem worth it, in the end.
And then I think of the parable : Jesus only got one out of forty leprous to come back and say thanks. So far I have a better success
Sancte Isidore ora pro nobis !
-----------------------------Sancte Isidore ora pro nobis !