The question is, if you believe AI will eventually catch up with all jobs that exist today, what skill could a significant portion of the population possibly have at that point, that the AIs would not also match?
Perhaps we'll have social constructs to provide jobs, similar to how people are willing to overpay for designer goods instead of copies (e.g. maybe 'hand-crafted' will take on a whole new meaning). But from a solely economic perspective, it seems like a matter of when, not if, an AI could more efficiently do most jobs.
Saying it's only about who pays is not really doing justice to the problems of the American medical system.
Companies can talk about cost reducing by cutting their workforce all they want, but I wonder how many stop and think how that would negatively impact the purchasing power of their former employees.
If somebody cherry-picks data to show 'some job growth' somewhere, and claims that disproves that automation is hurting jobs, then they haven't tried to get a job in manufacturing in the last decade; and forget any chance at promotion because there's no path from 'menial floor box-mover' to anything like engineering or management.
everyone will become a youtuber and other "personality" based jobs.
One other note, while I do believe that automation will cause massive job losses eventually, there exists an alternative narrative. Centered around the knowledge that: 'you don't know you don't know,' we just have no idea which jobs will emerge as a result of new industries powered by automation.
What does seem evident is that concentrated education at the beginning of an adult's life is unlikely to be enough in the future. Investment in education reform and adult education seems critical, whether or not automation causes large scale unemployment.