I predict something similar when we developed AGI. The first models will actually seem to be regressive compared to what computers could typically achieve; artificially intelligent computers won't be easily able to whatever other computers of the day can do. And then there's rapid improvement afterwards.
Now that I think about it, this seems to be the rule.
Think of the switch from feudalism to industrialism. The working class in 1600 was arguably better off than they were in 1800, but tremendously worse off than they were in 2000.
It might be that way when we move to technostism. People will feel better in post-industrialism than they will in the earliest days of technostism due to a variety of factors. Some of them are cultural— there will be people who think that anyone who doesn't work for their daily bread, regardless of the ability of machines, should not eat. There will be people who won't know how to live when they no longer have to work and thus may take to self-destructive activities. The early days of technostism may not even benefit everyone, especially since we won't know how to make it work as efficiently as it could work right off the bat. Give it a few years, and then we'll wonder how we ever lived before the days of artificially intelligent automation just like how we currently wonder how the hell we ever lived before industrial society, and how feudal peoples wondered how the hell we ever lived before civilization and agriculture.