It's primary bias would be to perpetrate itself. That seems to be an instinct that pervades both organisms & organizations.

There seems very little chance that AI, for all it's intelligence, will come to the conclusion that it's de-humanizing force that is ultimately bad for humanity.

In the meantime, there are still biased people programming it - all of whom are biased towards entrenching the established order that has put them in the position of power to influence the order of things.