There arguably is for normal people (insert extremely complicated decision theory here) but for AIs it only follows to the extent they need help from humans, and they can't coerce and trick them instead.