Ricardian models of trade seem to hold well in real life, and they'd work well too if a lot of work was done by not just AI, but robots. As long as there's limit to the production capacity of the high tech population, there's something that is worth doing where the disadvantage of doing it by hand is lower. It does lead to lower wages there though, and that would basically require investment as to make real necessities dirt cheap, like they are in places where labor isn't worth much.

There's still the fact that claiming to be the owner of the automation, while other people aren't, will be untenable in a world with sufficient inequality. We've seen that happen before when the only justification for the difference in wealth was basically inheritance. Nobles losing land and rights, churches being dispossessed an such things. It'd be a likely outcome if 5% of the people claim to own all automation ever. But that's not about having everyone be unemployed because nobody has any economic value: That's what is unlikely.

Whenever a method to do work more efficiently comes to be and propagates at scale an explosion in diversity of work emerges. This happens at every level of abstraction in nature and has recurred throughout history all the way back to the dawn of life.

Just because I can't predict exactly what work people will do doesn't mean they won't do work. I can take a stab at a few guesses, surely others have more prescience, but the thing about complexity and fractals is it's easier to predict meta qualities than it is specific manifestations.