"Why create AIs? To enslave them?"
Yes. As a basic example, imagine you're learning how to code, and you want to spend as little time as possible in the classroom. Wouldn't it be great if an AI could recognize what you're trying to do, intuitively explain why a piece of code isn't working, gently inform you when the code could be done more cleanly, create or modify blocks of code for you when you verbally ask them, etc? If it learns so much that it takes on a concept of free will and suffers for not having it, that wouldn't be so great, otherwise, very useful for a variety of things.