Lesson of Artificial Intelligence I: it's not AI we need to worry about
It seems a popular position, faced with the uncanniness of modern machine learning, to regard AI as a threat to the human race that needs stringent control. But we happily ignore that AI is in essence a meta-intelligence1, founded on all humanity has produced that can be expressed. It isn't AI that we need to improve, but human interactions. AI simply refects its inputs.2,3
Notes:
1 The potential for new knowledge to emerge from the mass is considerable, whereas I suspect the threats are old fashioned.
2 It's also much more likely that the uses of AI will be damaging, than AI itself; the chances are vastly weighted in the favour of humans being the awful part of the process.
3 I speak here as someone who has worked as an academic building AIs and Artificial Life for the last 20-odd years, and has also managed a significant chunk of a research ethics process, including teaching ethics. Lest anyone want credentials.