Your argument assumes that computers/AI/robots won't be as good as say a median human at any given job. I think that's quite a strong assumption