After I saw what Arcen Games did in AI wars, I wondered why more developers don't adopt the same approach, and try to design a game where the computer plays according to the same rules as the human player, only to fail again and again. Most classic games like Civ5, Master of Magic, Endless Legend, all Elemental games, etc. have AI that cannot really formulate any sensible plan on the strategic level. It can pretend to compete only after being given secret cheating bonuses, so it can mass troops faster than the player to compensate for its inability to utilize them properly. But the sense of challenge quickly turns sour when the player realizes that in a game where one side is allowed to cheat, all rewards are hollow. You do not win, because you overcame your opponent's strategy with better one, but because your opponent decided to conjure only 50 units out of thin air, instead of say 100, and it was still beatable. But it the AI would decide to cheat out 100 units, you would lose and could do nothing about it.
However, when it's explicitly said that the AI plays according to different (but transparent and clearly defined) rules, the game can become fair again.
I love the concept of AI progress in AI wars. You fight a pangalactic AI empire that controls virtually unlimited resources, and how much of it is dedicated to fight your tiny insurgency depends on how much you annoy the AI.
I hope we will see something similar in the Sorcerer King.