The real question would be is there a goal of the AI. The true superhuman AI, that is problematic, is when AI decides to learn something on its own. And I think that is a long road.