Its because, for perf reason, we have the AI and monster players all go at the same time (we multithread their turns rather than running them sequentially, so instead of waiting for 8 players to take 2 seconds each, all the players go at the same time and take a total of 2 seconds).

Because of this something player A and a monster player can both move at the same time and enter the same tile, even though the tile was empty to both of them.  It's not that the AI isn't getting attacked by monsters, or they are somehow at peace with them.  The AI gets attacked by monsters all the time.

The simotaneous moves could help the AI a little.  But you may occasionally see the opposite when the monster is moving to attack an AI player but that player is moving, in which case the battle still happens but it looks like the monster goes way out of its way to attack the AI player.

The real fix would be to remove the simotaneous turns, in which case the between turn time would be mutliplied by the amount of players in your game, and I'm not willing to pay that cost for this relatively minor issue.

Mr Kael, this function may be causing another related issue.  Fairly often when you select auto-battle, the AI unit moves before the battle resolves.  This can have the unfortunate side-effect of destroying your buildings.

Yes, if it had just two phases - one for Monsters and Ruffians/Denizens of the Wildlands, and one for all player AIs, then I agree that that would be the best compromise of all.  It would mean that the two monster factions could not attack each other easily, and player AIs might sometimes cross paths, but it would keep the player AIs from getting ignored by the Monster AIs.

From my understanding of the AI and multi-threading, is that all the calculations for the AI are being made during the players turn, this includes the monsters and all the other AI player. Suppose that you set-up the monsters turn then AI turn, then consider that the AI monsters would have to move and then the AI players would have to calculate their turn based on the AI monsters new positions. That couple of seconds would get extended longer than Derek suggests, even with that simple compromise.

Playing a late game of Civ shows this result of waiting for turn after turn. Mind you that the monsters are constantly populating the whole map nearly the entire game. The slow down for late game and early game is really really small compared to other 4x late games on equivalent size. I give major kudo's to the developers on this issue alone, as they have solved much of the lag problems. I would rather a unsightful, AI monster on same tile or AI monster ugly battle than what the compromise may produce in time delay.

Also, with that compromise in play, you may still find two warring factions of AI landing on the same tiles and other such things because of the same situation, so it wouldn't alleviate the problem entirely. I say they should work on other things than this issue at the moment.