Why do you think it's absurd? If we do eventually create an AGI that is significantly smarter than us in most domains, why is it that we should expect to be able to keep it under control and doing what we want it to?