In the current gaming market, graphics are a big deal. Everyone worries about how nice the new game will look; how pretty the explosions will be, how detailed peoples faces will be. However, something far more important than this, which is rarely emphasized, is the A.I. part of the game. A.I. requires significant computing power, as well as very good programmers.
A.I. is one of those things that can make a game truly great. Too often now the formula for success in an FPS is "Shoot at enemy. He jumps in cover. Wait for his head to come up. Shoot at enemy. Enemy Moves to cover. Rinse. Repeat." However, a clever designer could give them other options. Sneak around to another area is possible, or put just their gun out of cover and shoot wildly, or even just run the heck away after getting hit. This would add to the realism, and to the challenge of the game. Decision making in A.I. is often decent, but very few games reach the excellent mark.
Another part of A.I. that needs to be implemented is evolutionary learning, or learning to take different actions based on it's outcomes. In fact, this already exists to a degree (check paragraphs seventeen and eighteen). But it is very rarely used because it requires a large degree of research and understanding on the developers part. They must understand how people play the games, as too implement this formula correctly, and they must also account for a wide range of possible behaviors, creating animations for them, and variables, and etcetera (I'm still not an expert on how game programming works, beyond the ideas of cause and effect and such. Someday I'll get there). However, doing this would cause gamers to be challenged in ways that they have never been before.
Link used above, in case it won't work.
http://www.ign.com/articles/2013/04/08/whatever-happened-to-video-game-ai
A.I. is one of those things that can make a game truly great. Too often now the formula for success in an FPS is "Shoot at enemy. He jumps in cover. Wait for his head to come up. Shoot at enemy. Enemy Moves to cover. Rinse. Repeat." However, a clever designer could give them other options. Sneak around to another area is possible, or put just their gun out of cover and shoot wildly, or even just run the heck away after getting hit. This would add to the realism, and to the challenge of the game. Decision making in A.I. is often decent, but very few games reach the excellent mark.
Another part of A.I. that needs to be implemented is evolutionary learning, or learning to take different actions based on it's outcomes. In fact, this already exists to a degree (check paragraphs seventeen and eighteen). But it is very rarely used because it requires a large degree of research and understanding on the developers part. They must understand how people play the games, as too implement this formula correctly, and they must also account for a wide range of possible behaviors, creating animations for them, and variables, and etcetera (I'm still not an expert on how game programming works, beyond the ideas of cause and effect and such. Someday I'll get there). However, doing this would cause gamers to be challenged in ways that they have never been before.
Link used above, in case it won't work.
http://www.ign.com/articles/2013/04/08/whatever-happened-to-video-game-ai
No comments:
Post a Comment