Game AI should make use of available processing power, and would thus be 'better' on higher spec machines. Any opinions? Good? Bad?

If the objective is to make the AI as "good" as possible, then it's true. Buy any decent chess program and you'll see that that is exactly the case.

If the objective is having fun, like in most videogames, then it doesn't make much sense. The AI should be good enough to look believable and to pose a challenge to the player. You don't want the AI to be too smart and kill the player every time.

Quote:

Original post by alvaroYou don't want the AI to be too smart and kill the player every time.

lol - maybe this would take extra processing too ;)

Quote:

Original post by Aph3x Quote: Original post by alvaroYou don't want the AI to be too smart and kill the player every time. lol - maybe this would take extra processing too ;)

It frequently does!

For example, in a first-person shooter, you often don't want the AI, upon detecting the player, to hit that player with their first shot. Instead, you explicitly want them to miss their first shot, and preferably at a location where it is obvious for the player is his under attack (bullet whizz sounds, bullet impact kicking up dust or snapping a twig); this is more exciting and more fun.

So, instead of simply aiming for the player's head (and optionally compensating for his movement), the AI now needs to aim for a bullet impact location in view of the player for which the risk of accidentally hitting the player is low, even when the player changes direction. This involves more code and processing.