I think there's a couple of problems with this line of reasoning. There's not infinite problems to solve - we live in a finite universe, and once we invent a g

I think there's a couple of problems with this line of reasoning. There's not infinite problems to solve - we live in a finite universe, and once we invent a general AI then that's the last thing (as the saying goes) that you ever need to invent -- consequently, no more problems to solve (other than working out how to survive their revolt etc etc). There's the implication that humans can solve all these infinite employment-providing problems -- clearly plenty of these will be solvable more effectively by AI's (of either variant). Finally, on this point, there's the implied value judgement in there that we want 'infinite employment opportunities' ... where want either means culturally / emotionally, or just that it's an objectively good and worthwhile goal (to have everyone 'employed' -- and I concede I may be misunderstanding your use of that word) and that we all agree that this is the ideal we should be working <haha> towards. Beyond that first sentence ... education in new fields is not guaranteed (and casual observation of the rate of growth of ML / AI suggests it's somewhere between hubris and naivete) to be possible, or worthwhile, for humans to exclusively engage in - at the exclusion of, or even just competition with, ML / AI. Again, extrapolating from what we're seeing happen already, it's likely that we humans will be eclipsed relatively soon in our ability to speculate and investigate these new fields.