This topic is locked from further discussion.
it depends on how it's implimented and the ai used. sure humans can learn to predict a bots actions, but the bot can be programmed to try other actions at random and use those actions with the best results. a memory could be added to keep track of what humans do. the ai could also auto adjust the dificulty to stay competative.
unreal is a good example. in ut3 it should be hard to tell the bots from human players. on the pc you could creat new ai's. you could download new ones for the previouse uts. and they aren't anti bot, all the enemies in single player are basically bots. they just dont have them in multiplayer. and ai can learn. turn ten uses a learning neral net for it's ai. i think they freeze it (static, non learning) before they put it out at retail. but it does learn in the studio, but if it makes a mistake or does something random which has posative results it will repeat what it did. it wasn't taught to break check (touch the brakes slightly to make a drafting car think they are going to hit you causing them to slow down excessivly or swerve off line), but it does it. ai today is better than it used to be. it's getting more adaptive than just a scripted sequence.
Please Log In to post.
Log in to comment