This clearly doesn't apply to everyone, but I think it stands true for most.
I'm not a troll and I am in no way trying to stirr things up, but I feel like the truth should be said.
If the game had a decent AI, 90% of the players couldn't enjoy it.
Shouldn't 'the truth' be supported by some kind of reasoning? No one is talking about neural nets, which require a lot of extra processing that would delay turn times apart from anything else - they're talking about AIs that can use the game's basic systems. Difficulty level has nothing to do with it, since difficulty in Civ games has always been built around bonuses given to AI civs and/or penalties added to the human player rather than any differences in AI behaviour.
Deity is too easy, but if the AI were able to use the bonuses it is given extra bonuses could fix that without affecting play at lower levels. I don't know if Civ VI AI has the same issue (it doesn't seem to hoard gold nearly as much), but in Civ V one of the AI's bonuses was extra gold - but it didn't know how to spend it, so you just had a bunch of AIs with giant treasuries that the human player could exploit with trade deals. For bonuses to have a meaningful impact on AI difficulty the AI first has to be capable of using those bonuses.
Some complaints about the AI are idiotic, for sure. People who use all the available exploits, like choppiing overflow combined with wall/naval production cards, and then complain the game is too easy I have no patience for at all. You can't reasonably resort to cheat codes and then complain the AI can't keep up - and frankly I don't see the point anyway. I find Civ VI too easy on Deity, but it is at least a moderate challenge played fair.
Similarly, people who insist on wiping out the AI civs and complain the AI can't match them militaristically have unrealistic expectations.
None of that ignores significant functional problems in the AI's simple ability to use basic game systems.
- AI civs consistently fail emergencies against other AIs, because (and I've observed this while participating alongside AIs in emergencies) they'll mill around and kill enemy units - and can do so quite effectively - but will almost never attempt to attack the target city.
- AIs do not use governors deliberately to manage loyalty - they'll routinely put a governor in a city for a few turns, then move it out again seemingly at random, and have trouble with the loyalty system in general. They frequently forward-settle cities they can't possibly keep, and although their city placement is generally okay (they'll often be a tile off the optimal spot, but close) they will still insist on sending out settlers late in the game and colonising any unused land, even if it's in snow.
- AIs use air forces and nukes extremely rarely, even when the game goes well past the point when they develop those techs.
- In wars past the early game the AI will often not attack, and when it does it's reliant on the large number of units it has access to at higher difficulties to do so effectively. Which however does not stop them declaring war, even against civs that are not close neighbours.
- The AI still repairs pillaged spaceports infrequently and seems to protect them from spies more or less at random, without favouring ones with an ongoing project. Since the AI loves blowing up spaceports with spies, even without human interference this can stall their ability to complete a science victory.
- AIs do not seem to value Great People - in previous patches they seemed to faith or gold buy them sometimes, but this behaviour seems to have altogether stopped.
- AIs love conquering city states instead of using their bonuses, and do so right at the start - a waste of resources that is only likely to slow them, especially on higher difficulties where the extra settlers they get make an extra early city less valuable.
- The AI still struggles to use garrisoned units correctly. It's somewhat inconsistent whether it garrisons with ranged or melee, and if garrisoning with melee the unit inside the city will almost never attack adjacent enemies.
- I'm not sure quite what the fault is, though it's no doubt some combination of the above plus the AI's tendency to waste resources on unnecessary wars with other AIs, late-game settlers, Holy Sites in most cities, excess spaceports and so on, but it is simply far too easy to overtake the AI for peaceful victory conditions. In my last game I checked - I was still the last of 11 (Norway had been defeated) AIs in science by turn 235. I won a science victory 70 turns later, through nothing more complex than pushing campus/industrial zone projects to get the space race GPs, and despite missing on Ruhr Valley - and, needless to say, using no overflow tricks (in fact I ended up wasting part of Korolev's boost since I didn't have the techs to chain two space race projects when I used him). This should not happen on Deity.