I gave a lecture at the University of Toronto last year with a similar point to what the speaker was getting at. Namely, we should be focusing on augmenting human intelligence, not chasing some sort of future robot overlords. See https://adamdrake.com/novel-results-considered-harmful.html This was the approach Douglas Englebart took in the lead up to the Mother of All Demos (MOAD) and it's unfortunate people have for…
> we should be focusing on augmenting human intelligence Who’s we? I can throw machine learning at practical real world problems right now. Augmenting human intelligence? I’m not aware of any work in this area.
Great! We (technologists) should be using machine learning approaches to practical real world problems, particularly those problems whose solutions allow humans to be more productive. That's what I mean by augmenting human intelligence/capabilities.