You've seen scholars in residence pass?
I've literally never seen it pass unless I had enough votes alone to force it through alone.
To be fair, it was with in an AuI test game, and AuI changes a lot of the scoring behind world congress proposals, one of them being a bonus or penalty to scholars in residence's score based on how well the AI is doing tech-wise in absolute terms (eg. if you're halfway through the game, you should be at least halfway through the tech tree).
A question then for those who have viewed/run multiple simulations. What principals can be gleaned from these simulations about which AI Civs are generally stronger (possibly just due to how well the AI can use their advantages), what makes a strong AI Civ, what AI attitudes or other conditions contribute to overall AI performance? Are there specific AI Civs/attitudes that tend to always dominate when there is no human player around?
Most of it relies on how well the civ's UA, UU, and UB fits into how the AI works. For example, Poland's, Japan's, and Greece's UAs can be used pretty well, while Maya's, Sweden's, and Austria's UA are used fairly poorly. UU's that are essentially just stronger versions of existing units, eg. Mohawk Warriors, Longbowmen, or Musketeers, work much better than UU's that change a unit's role, eg. Camel Archers, Siege Towers, or Battering Rams, or UU's that depend on you using the UU's unique promotions, eg. Chukonus or Janissaries. The AI is programmed to make use of some "weird" uniques, eg. Nau's trade ability or Celts leaving unimproved forests that give their cities faith alone, while others are completely ignored, eg. Feitorias or Samurai and Legions being able to build certain improvements.
The usual rule of thumb is that the more "ordinary" the civ's uniques, ie. the more it falls in line with the AI's algorithms (simple boosts to stats that don't require special strategies are a good bet), the better the AI fares when using them.
Essentially, all AIs actually have the exact same AI. The "flavors" system helps a tiny bit, but can't account for the specifics of each Civ. The one thing that generic AI is consistently good at is building an economy via expansion and an army. Any Civ that doesn't thrive when doing one of these two things will be weak.
There are quite a few exceptions, but there's no logic behind which bonuses are accounted for and which ones are ignored. For example, the Huns' bonus for razing speed is factored into the AI's decision what to do with captured cities, the Netherlands' happiness retention bonus is factored into their dealing/diplomacy algorithms, and Indonesia's luxury bonus is factored into the AI's decision of where to settle cities, but Sweden's DoF affinity is completely ignored, Austria's UA is only ever used as a passive for acquiring the CS allies they already have (they don't actively look to acquire CS allies to purchase), and France's UA is ignored when deciding where Great Works are shuffled each turn to maximize theming bonuses.
What about a map that was all of the warmongers together in a death match? Shaka vs Bismark vs etc etc etc.
The AI for Assyria is probably as bad as the AI for huns, eh? I've virtually never seen seige towers get built by the AI even though they are awesome at taking cities before gunpowder.
I actually think that AIs who are aggressive builders would work better than pure warmongers: Russia, Poland, and Greece come to mind.
As for Assyria and Huns not using their siege UU's properly, it's because they are still marked as non-siege, melee units, because there is no handling of units marked as siege units who are not ranged.