I would prefer to play the most 'intelligent' setting for the AI with small boosts, others like to play the hard AI with big buffs. Therefore the options should be separate so each player can pick the setting they find to be the most fun.
It would be nice if the AI could be "intelligent" (an "intelligent" setting) and play the game "intelligently" and provide a challenge in a 1:1 way (i.e. no buffering or advantages of any kind for the AI: what the human get is what the AI gets). Unfortunately, I simply don't believe this is possible to do in a complex strategy game like EWOM (and this is *any* complex computer strategy game). I have no doubt that it can be "improved", but not made "challenging" on the 1:1 level.
I'd be with you 100% if it was possible, it's just not. "Artificial" intelligence is just that, it's "artificial" not "real". Maybe in the future AI will improve even to mimic human "intuition," "foresight," "planning," and "strategizing" which are essential to good 4x strategy game play, but I think we are far-removed from that time yet.
I'm just saying that even if the computer plays fairly, it's not quite an equal footing to the human; because they can do certain things that aren't technically cheating, but do give them an unfair advantage.
True, if you buffer an AI (by choosing a more difficult "difficulty" setting, or slider setting as the other poster suggested), you'll be giving the computer an unfair advantage (as the human player has an unfair advantage with a "human" brain). But, what players are doing is essentially trading the AI an increased advantage (I don't personally see this as "unfair") in return for the ability to play a more challenging & fun game.
If a 1:1 game turns out to be just an AI bash (which is not enjoyable to me), I choose to "buffer" the AI (or increase its advantage) in order to get (in return) a more challenging game (which = fun). The problem is, right now, even on ridiculous the AI is still a walkover. I hope they will "fix" this, and in fact SD has been improving the AI a little bit with each patch which is a positive thing.
Let's see how version 1.1 turns out.