Yeah it's a matter of time, I've said it many times before but here's how I see it playing out (warning schizoposting):

Constitutional AI gives people the opportunity to 'create' bosses with publicly available constitutions, or plain-english codes of behavior. There can be contractual obligations not to change the constitution for five years, workers get a vote, etc, the point is it can be deliberately programmed and open sourced for 1. Maximum trustworthiness, and 2. Maximum employee benefit, meaning all the money that would ordinarily be the CEOs would go to the working class.

Since AI CEOs have a built in advantage of being capable of being deliberately programmed not to be greedy, which there is HUGE incentive and ability to do so, that gives them huge advantages over human adversaries. Humans will be held back by their own greed because the money isn't efficiently being distributed through the company.

Human CEOs will be caught in a bind, the only way they can survive as CEOs is not to be greedy but if they aren't greedy that defeats the point. Any competing AI CEOs that try to be greedy will simply not be able to keep up, it's a huge advantage because it means for now they will be taking on some or all of the human workers displaced by human CEOs automation.

The only point of disruption is that human CEOs have a lot of assets, so even though AI CEOs will have the workers, will it matter if human CEOs are the ones with all the means of effective production, like factories? That's the only part I can't get my head around yet.