I'm not really sure if I buy the argument that AI will equalize society. A more realistic goal would be reducing poverty and keeping the Gini coefficient within

I'm not really sure if I buy the argument that AI will equalize society. A more realistic goal would be reducing poverty and keeping the Gini coefficient within reasonable bounds. I don't think this is necessarily incompatible with your hypothetical "robot butlers and oligarchs" picture. Just because a few people get ultra-rich from a new technological development doesn't mean that everyone else suffers. China has seen an increase in inequality over the past few decades, but has also significantly reduced the number of people living in extreme poverty. If breakthroughs in AI increased productivity to such a degree that we could give everyone a basic income, raise the minimum standard of living, and keep the Gini coefficient in check, that would be a good outcome for everybody. There is definitely a self-serving aspect to tech millionaires and billionaires advocating that we should accelerate the development of the technology they happen to be working on. But I also think there are principled arguments for accelerating economic growth. We have a lower proportion of people living in poverty today than any other time in history, and that's largely because of the economic growth caused by the technological revolutions of the 20th century.