In this thread, I'll list 9 reasons why AI factions are weak in late game, over human players, in no specific order, while avoiding obvious stuff like battle AI or similar.

1) Rush buy economy

In this game, rush buy economy is awesome. Getting tech that gives discount for rush buys, and two levels of economy and science/industry policies will make rush buying especially effective for cheaper buildings (pretty much anything with base industry cost below 1200). When this gets set up correctly, and money starts rolling, only real limit in development are the strategic resources, and not industry or dust price of buildings.

AI factions are not well aware of this, making this strategy especially effective to catch up to AI factions, on higher difficulties.

2) Lower tier techs are often better then higher tier techs

If there is tech that improves the city with new building that is cheaper then another tech that does the similar thing, you really want to get the one with cheaper building. There are also techs that give powerful unique bonuses, not available in higher tiers (like rush buy bonus), as well as those that give faction specific units. Only exceptions are very situational techs, like bonus on rivers and stuff.

Unfortunately, AI factions are not aware of this. As matter of fact, it looks like they just research minimum needed amount of techs in the era, and then hop to research techs in next are, never looking back. This is horrible strategy, since often that means key economy buildings are left out, or that faction unit techs are never researched.

3) Market resource trading is awesome

And AI factions mostly just sell the excess resources and rarely buy them, due to AI bonuses. Which also means more stuff to buy for human players, which is great for building up tall empire on higher difficulties, when getting outsettled by AI factions.

4) Stockpiles are awesome

Stockpile price in the market is mostly fixed. But two techs upgrade 100 point stockpiles, first to 600, and then to 1600.
With those two techs you can buy industry stockpile worth 1600 for just 300-400 gold. And two of these is enough to create new 8-unit stack from scratch. Get six, and you got 3 8-unit stack armies that can go in war of conquest.

5) Stockpiles are awful ;-)

Building stockpiles, especially industry stockpiles is awful for most the game, since using economy to rush production is much cheaper, without loss of 1/3 of the spent in dusty. The only exception is building end game wonder. Food stockpiles are not great either, due to diminishing returns, and science stockpiles only make sense if you really really not need to produce anything (I would still add more military units).

Still, AI is not aware of this, and will spend a lot of time queuing stockpiles instead building, for example, more units, if there is nothing else to build.

6) Unit upgrade system is not flexible enough to cater to strategic resource scarcity

The problem is that units created with design from one unit template can only be updated to other units using design with same template. So you can not have one design that is using dust only equipment and another with strategic resource, with option to upgrade first one units to another.

Human player can easily work around it. By creating new design, upgrading only units they want and then reverting design to the old one.

AI factions can not use such finesse. Usually what will happen is that if their design is using some scarce strategics is that AI faction will just stop making those units, if there is lack of those resources, instead still making more units with humble dust redesign.

7) Dust equipment unit designs are very cost effective

Most of the armor effects that require strategic resources are very minor upgrade of dust/iron equipment. Weapon bonuses are good, but often very expensive, especially later tiers. On the other hand, dust accessories are just awesome, which makes dust-only unit with health and regeneration boost pretty viable against unit designs that use strategics.

By sparing strategics resources using dust unit designs, human player will have enough of those resources for key buildings and improve their economy, making them stronger on the long run.

8) AI is very averse to food focus in their economy

This is not bad on first glance. High food focus is not great strategy, since extra population does not contribute much. But biggest strength of the extra population is that it is prerequisite for the districts, which are really powerful. Thus, in middle game, human player faction starts outpacing them with bigger population and thus more (rush)built districts.



Hope this list was fun read. Most of this stuff was observed by playing couple of serious difficulty games which always started with AI outsettling me, forcing me two play tall strategy and then me outpacing them in economy and military in middle game, making conquest of nearby factions possible, which makes going for any victory condition doable.

That is pretty much always the gambit with AI, how do you give them enough of an edge that they can give a stronger challenge but allow them to exploit and push their advantage? It's Sysiphus's rock all over... but you have to imagine Wilbefast is happy. :D