I am going a run a test myself. I am making a series about the conquest of America with a dual purpose: to guide a player that doesn't know what to do and wants to play historically, and at the same time to help the AI achieve a more historical result.
For example, the first event fires in 1492 and hints that la Hispaniola is the first place colonized. The following fires between 1501-1507 if la Hispaniola colonization is underway, and says that Cuba goes next, all the while giving a little help, mostly intended for the AI. Around 1519 an event will direct the player to Tlaxcala. If he conquers it, it will fire another one telling him to go to the capital and giving him some Tlaxcalan allies. We know what happens then. But a parallel AI only event will directly cede Tlaxcala to Spain in 1519 if discovered, together with lots of indian recruits and a fancy war declaration a few months later. That ought to do the trick 90% of the time, given that the AI is above level 5 always.
This should keep it still interesting to the human player, and provide the AI with enough help to perform historically. The info in the events will substitute the cores for the player as hints. The AI will basically be forced to go ahead and should be expected to accomplish in about 95% of the times.
I will be posting them in the Iberian thread if you want to follow their development and contribute to it.