The problem with a static AI is that in the very best case it's just about as good as you without exploits, and that way it will be down to random luck (close resources, a roll that keeps your lvl 4 unit alive in that one battle, etc) Play that then long enough and that will seem to be bad. Humans opponents > AI up till now (unless you play for different reasons (story, figuring out the game etc)).
Why can the AI only be as good as the player without cheating? Most chess and backgammon AIs will beat any human player and they achieve this without luck. Now granted, those are relatively simple games (in terms of the rules, not the depth), but they are also far beyond what most humans are capable of, so they could do worse and still be better.
Have you played AI War? This game probably has the strongest gaming AI right now, and the one in GalCiv2 was very excellent, too. I think a lot can be done in the AI field for games, but it takes effort, expertise and a budget. If a game has multiplayer, the AI tends to get even less attention since "real people offer the best challenge" is a convenient excuse. In case of Elemental, Brad is going to spend significant time working just on the AI, which to me illustrates how complex a task this is.
AIs that cheat (i.e. get advantages such as information or resources) and therefore play by different rules than the players are something I never liked in games. It bugged me in AoW, too. (Still an excellent game, one of the best.)
Hmm you quoted a part of my text and then seemed not to have read the rest, or at least misunderstood.
I am sorry for that, let me try and clarify:
As I said: the AI is static, meaning that it doesn't change tactics (even if it has a random generator choosing between different styles that's still static, it cannot learn, grow, develop, try wild things not already in the code). So, in the best case scenario it is just at human level, meaning that you win 50% of the time. (I call this best because winning 90-100% of the time seems to me to get boring or annoying (we are talking years of gaming anyway, I am not talking about when the game is still fresh and new or when you are playing it for the storyline or such) and losing 90-100% of the time as well.) Now, when it is at that best level (basically as good as you 50% of the time, then the only possible way for it to remain that way (again, we're talking long term here), is there to be no further player learning possible that makes the player better and that would then shift the balance to an undesired percentage (remember the AI is static so it cannot evolve along (yet, could be possible in the future, and yes, difficulty levels can simulate that for a while)). So if you win 50% of the time, you have maxed out the strategy/tactics and cannot learn anymore, then the biggest deciding factor of who wins is no longer skill, but random things, like resources starting far or near, or a lucky lucky shot in a certain battle. In the end, those random deciding factors get known to you and it will turn boring again. That's why I say human opponents always are better then AI ones, till now anyway.
I hope this made it easier to understand what I mean.
EDIT: and about AI exploits: I don't like them either, so that's why I say best case scenario is Ai not using exploits and beating you 50% of the time (roughly of course, say in a +- 15% range). One could also argue AI cheats do not matter if it wins 50% of the time against you, but i just think without cheating is preferrable, cleaner, better coded, feels better as a player.