There is a lot of reasons why automated companies pay more but it is usually to do with augmentation and the remaining skills required being higher on the skills pyramid and more scarce. Efficiency gains usually cause the next bottleneck to be hit requiring more skill and/or experience to fix (higher hanging fruit).
Usually the amount of pay increase no way compensates for the total pay lost across all workers - that's one of the incentives to continue with the efficiency gain of course. If it happens too fast only a select few will have jobs because the demand won't have time to match the new supply.
All the above assumes you need greater levels of intelligence to deal with the next bottleneck and that intelligence comes from humans as its only source. If AI is more intelligent than humans then this effect - known for most of human history - becomes an invalid assumption of the economic model; bottlenecks and the ability to solve them goes to the highest/cheapest intelligence.
IMV should AI do this it would invalidate most economic system assumptions including ones in capitalism, socialism, etc. My base case isn't a positive one sadly of what comes out on the other side.
I suspect we will invent bottlenecks for ourselves. There is already so much bureaucratic nonsense involved in knowledge work. For example, in a lot of states for a lot of purposes, a document is invalid if it is not notarized. The notary doesn't really add value in a market sense, we have just decided that you need to go swear in front of another person that you are telling the truth or that the document is genuine. Is it really that much of a stretch to imagine that there will be a human review requirement for lots of things in the future, even if completely irrelevant (or actively detrimental) to the actual work?
What's your base case for what comes out on the other side?