I bring this up because of a YouTube video I've seen recently from Computerphile about the Three Laws. If you don't know what the three laws are...why are you here reading this? Anyway. The expert in the video explains so well how ridiculous the three laws are when talking about a.i. I think his point explains why, though I'm not an a.i. researcher, I haven't considered the three laws. When you're thinking about getting a machine to learn to solve a problem, it's so obvious that the machine has to follow such a deterministic line, no matter how curvy or straight, it just can't solve for every possible set of data...so the idea of expecting machine learning to even understand the first law is so obviously out of the realm of possible at the moment that it truly isn't worth considering the three laws. And as the speaker points out, Asimov created the laws to write stories where the three laws are violated in some way or other. Really wish I had thought to write about this before the video, but I clearly don't have enough people commenting on here to get a few commentors who would ask such a question.
So what is a human? What is harm? Write your answers on a sheet of paper, and see if someone can think their way around the laws given your definition.