IMO management positions are mostly lobbied for/created by try-hard social climbers, at least initially. "We're taking a lot of X work, maybe I should lead a team to deal with that?" "Y has so many reports, maybe I can form a subgroup to help with that?" Creating new positions for people who want to be more important than they are right now is the main mechanism by which private orgs expand.

Doing this is considered proof that this person is a natural leader who steps up to solve organizational problems and get things done. You can guess why this leads to many many layers of management mostly just having meetings with each other, and a confused bottom layer of people who have to use this deliberately broken human telephone to communicate with their real ultimate bosses, the owners.

I have a suspicion managers will become redundant sooner than tech workers, although certain big CEOs love to try to say otherwise... (wonder why...).

An (good) AI manager is far more efficient than any human manager, and doesn't need to resort to this tiered system. In theory, they are far faster any any human manager too, meaning the company can scale around them without any issue.

Maybe you still have a board that reviews decisions at a high level, and an office of human manager cogs that can review the individual AI decisions, but then your company structure can become such that a corp of 1k+ individuals can _directly communicate with their customer(s)_

Now, of course I'm not going to pretend that this won't come with its own share of issues, but that's what the "manager cogs" are for...

>and a confused bottom layer of people who have to use this deliberately broken human telephone to communicate with their real ultimate bosses, the owners.

Which just feels like efficiency if you're the owner: less people reaching out to you with problems!