Re: Wanted: Cheating AI
For my part, I would like to be able to determine the ways in which the AI can cheat. A headstart in resources, racial points, planets, etc. ? Yes, I give that already to the AIs, however, it is me who decides. A better use of all available information (in a way humans could not use it)? Yes, yes, yes, that should be exploited. Let the AI e.g. simulate EVERY major battle before committing its forces. However, the simulation should be based on known/"guessed" designs because giving the AI information which it cannot have is a kind of cheating which I do not like (however, the AI could spend more intelligence points on knowing designs for example).
Nevertheless, though I am also no programmer, I think there is still ample space for AI improvement. Target evaluation (what do I attack, when do I abort my campaign, etc...)should be possible to do. Maybe some behaviour must be "hardcoded" into the AI (with some stochastic element) until a better solution is available.
|