We here in my house actually just the other day had a conversation about AI, because apparently there are a lot of conversations going on in my husband's industry about how they're all going to be replaced by machine-learning soon. Which is interesting, because I would not have picked the legal industry as a front runner for that, but there you go.
Anyway, Mr Dee is not troubled by being replaced by a computer, because he figures by the time AI gets that smart, and I quote, "we'll have other problems, and they will be called SkyNet".
I guess my point here is: don't dismiss Terminator-type stories out of hand. They are a fundamental subset of stories-where-AI-doesn't-follow-Asimov's-Laws.
So I guess my real point here is: I'd recommend starting with reading Asimov, or reading about Asimov's
Three Laws of Robotics. It's certainly an active concern of my husband's that the world take some steps to having a central regulatory/ethical consideration about their observation in the developing field of AI/machine-learning.