@PhilBowles
I rarely ever see the AI pillage in CiV either. Whereas in CIV sometimes I felt the AI's whole goal in warfare was to pillage. It's weird and I don't get it. The AI also does not usually destroy citadels (I've never seen it but others say it happens occasionally) which is dumb because occupation doesn't flip ownership.
I've seen it a few times, but it seems to be largely at random - Attila had a hard time getting to one of my cities through the adjacent citadel; in fact I'd even left the city itself undefended since he really couldn't work out how to get past it. He eventually hit on pillaging it, but once I'd driven him off and repaired it, he never subsequently did so again, despite having several later opportunities. I'm not sure how difficult it would be to develop an algorithm that tells the AI that, if by chance it has greater success doing something than it had before, repeat that behaviour in this and future games - nothing as complex as a full-blown neural net or any particularly developed ability to learn, just a priority list with some kind of scoring system that will make certain behaviours rise in priority if they prove successful.
@BrokTheFanatic
I also dislike the diplomatic victory in CiV. Not only does the UN not do embargoes and the like from CIV, but the whole way someone can win by showering gold on City States at the end just feels cheap. There have been many times where I could build the UN and win the game but I don't because it just feels...lame. I still leave the diplo VC open just to see if the AI would ever go for it. They don't in my games.
They rarely build the UN, but if you have it they'll try their hardest to go for a win through the UN - but only by trying to grab city-states, not by winning favour with other civs, and even your closest competitor will vote for you just because he likes you more than anyone else. I've taken to playing a lot of games recently where I play the neutral party and just end up farming everyone else's votes.
I've only once had to resort to gold-buying CSes to wear down an AI rival going for diplo, and it's also the closest I came to losing through diplo (indeed I did lose by 1 vote but reloaded). It can work, but it's not the most reliable or efficient way to do so and is easy for the AI to counter by seizing your allies and declaring war (a tactic it knows) - it's less easy (due to AI difficulty taking cities) for the AI to counter by invading your city-states, but it can happen. I mentioned one game before in which the deciding vote ended up going to England (which I'd liberated early in the game) because I'd lost Jakarta to a Cathaginian invasion.
Besides which, of course, making enough money to have the reserves to outbuy all competing AIs is a challenge in its own right, so it's not clear why a "make lots of gold" victory would be any more "lame" than a "make lots of culture" victory.
Diplo is by far my favourite victory condition, all the moreso since G&K - there are more varied approaches to winning it, it brings in more aspects of gameplay to succeed than just a tech rush to the end, and post-G&K your relations with other civs can be as or more important than your relations with CSes.