Questions regarding AI, and a side question regarding spells

[PsiNorm]

What factors does the computer think about when it chooses to research tech?

For my L5R mod I want to have a branch of the tech tree be about corruption. It will provide strong units, but at a cost of producing pollution and making people unhappy.

If I do make that branch will the the computer research it and build away? I'd be great if there was a way to set a percent chance that a civ would go that route, but that sounds too good to be true. I want some civs to use it and others not to (but make it available to all). Can the AI see the downside and choose that that's not for them (and at other times feel it's necessary to win)? Big order, I know.

I read in another post that you have to have all the terraform options checked if you want the AI to use a certian one. True? I want to have a unit that can clean up the taint (pollution) corrutpion causes and it'd be silly to have those guys running around building roads. Any work arounds?

Ok, regarding spells. Some units will have spells built in (like bombard or ranged attacks), and others will be carriers that can load a spell (a la missiles). Any ideas on other cool spell efects? At the moment all I see are a few spell types that simply progress in stregnth. I'd like as many "types" as possible.

Thanks,
PsiNorm

 

[pdescobar]

I can't speak to AI tech-choosing because I've never really understood what the AI considers an advantage or not. I am cynical enough to believe that they'll research about anything you put up if it results in some strong units, regardless of the downside. Only by testing can you really gain that kind of feedback, though.

You must have all the terraform options checked in order to give something the "terraform" AI strategy. You could, in theory, uncheck other options after doing so (Terraform remains checked but greys out), however I have no idea whether or not the AI will use such a unit properly or not. That probably wouldn't be too difficult to test.