If the AI is properly designed/constructed with the goal of servicing the flesh-bag population, that is what it would do. It's a fundamental error to assume tha

If the AI is properly designed/constructed with the goal of servicing the flesh-bag population, that is what it would do. It's a fundamental error to assume that AIs will just be artificial human minds (with regular human goals, motivations and aversions). "Properly designing" the AI is still an open problem however, which is known as the "alignment problem".