Academic AI vs. Games AI

• Here's some actual content instead of an overview
• Part of the reading today (see conclusion ) discusses this.
• Buckland is right that AI is divided into Hard AI vs soft AI. Hard AI trys to do things the way humans do, and soft AI is happy to do anything that works.
• I don't think he's right about optimality. He states that most AI doesn't care about time of computation, but are just interested in getting the best (optimal) solution. I think most of it is about satisficing, getting a good solution. Most has to deal with computational limitations.
• That said, Buckland is right in that most AI is not required to deal with real world interaction like games AI. It can just toddle off for a while and the user will wait (say minutes) for a solution.
• The time limitation does make things more complex.
• Your agent can't spend time calculating where best to shoot when it needs to be doing something else. A good guess will work.
• Let me also give a little whinge. Every game says it has AI. Often it is just some hack that I wouldn't call AI. They use AI as an advertising gimic.
• The hack might be a good idea but hack probably doesn't sell well.