I believe hard AI is going to be a lot more problematic to map out than we currently might think. I believe we're in for a push-back against overly-intrusive computer monitoring, which should slow down AI development. I also believe that "intelligent", "kind", "moral", and "useful" may also be much more mutually orthogonal than currently believed. Finally, mankind really needs to work through the period where mid-level AI could either exterminate us or love us into extinction. That's going to take some time to work out.

But it's all just guesses, of course.