All environments are not created the same, and therefore not equal. If you feed 100 environments that are nearly identical to each other but favor A, and 4 other environments that all favor B but are all dissimilar to each other, then although B is more generalized, A would still win.

Since you have no formal way to compare environments to each other, you can't prevent this from happening. Therefore you just pushed the subjectively as to which AI is smarter to which environments are chosen by the user to run against.