I hate to say this but if you find the final 1.1 AI "pitiful" then I think we're just up the creek.
We're nearing the end of what can reasonably be done without massive cheating to give the AI lots of extra bonuses and such.
I tend to think that most people who are winning at the highest levels are probably making use of some exploit.
I can't beat the AI at higher levels than "Gifted" anymore with 1.1 unless I set the starting conditions in just the right way and make use of very specific strategies (i.e. massive surprise attacks, diplomacy manipulation, etc.).
At a certain point, there's only so much I'm going to be able to do as a developer to make the AI "smart" without wrecking it for the 90%+ of people who find the AI challenging and want it to be "fun". Not to mention, there's a budgetary issue.
I suspect the AI plays best at medium sized galaxies with a reasonable amount of planets with 4 or 5 opponents playing.
I think those who find it "easy" are also likely playing at say a gigantic galaxy with rare worlds where it's really hard for the AI to make the most of its native computing advantages.
But we're nearing the end of the line of what I'm going to be able to do. The AI will continue to improve but there's not going to be any sort of earth shattering changes beyond 1.1 in terms of AI.