Why an AI for Outwitters is hard to do

[Kamikaze28]

(05-04-2012 04:59 PM)Solan Wrote:  I suppose that as far as programming the ai to predict uncertainty, people can't really predict that either, so you could program it to do what a human might do; try to maneuver in your own strategic way. for example, the ai could predict where the enemy units could move to a certain extent, so it could try to do it's own strategy, like surround a possible area the enemy is in, spreading out it's units, or retreating to reinforcements. An A.I. for a game like this could just have a few sets of default go-to strategies to use in the case that it has no idea what the enemy is doing... at least that's what i think. Wanting to be a video game programmer myself (although i have no experience programming so i can't even begin to do so ), i feel like making an A.I. might be the hardest part of making the game.

If the exploration of the game state works as intended, you won't actually have to define these go-to strategies, they'll come as a bonus because they are the most promising actions to take given the current state.
I would also disagree with your statement, that humans don't keep track of uncertainties. Novice players may not do so, advanced players may do it subconciously and high skilled players will probably do it conciously. It's an integral part of being human to handle uncertainties, we've evolved this ability to survive, actually.
And to your last point: I completely agree that AI is probably the hardest part in a game like Outwitters, which is why OML completely omitted it in favor of the league system and matchmaking.