Tell us the result afterwards, ya know.
I've ran 5 tests so far, each with a randomly-specified player plus 34 randomly-selected AIs. I was guessing about a 35% chance of the bug occurring in each of these games - but never got a duplicated AI. I'm giving up, and considering my previous "bug" observation to just be a random low-occurrence fluke that does not need a complicated fix. If anybody runs into this same fluke, they can easily use WB to check the player pull-down list for same-name AIs. If starting a new game, regenerate a new game. If continuing a long-term existing game and noticing this fluke, use WB to find and erase all cities settled by one of the duplicated AIs.
My apologies to all for previously calling this out as a bug that needs investigation and fixing.
Other things I accidentally learned:
(1) At the current code level, game start-up limits the total number of players including AIs to 40 (but only 34 fully visible due to the 6 automatically-generated AI slots at the end of the list (The Green Man, another Green Man, another Green Man, a Barbarian, Neanderthals, and another Barbarian). This seems to happen whether or not Barbarian players are specified On or Off.
(2) Each new game start-up with this many AIs takes a very very long time on my old gutless laptop, and also if I try playing a few turns.