It's not an AI thing. It's an organization thing.
Two big breakthroughs were needed to create organizations bigger than one person could own and run - the corporation, and the org chart.
Daniel McCallum invented modern business organization in 1854. [1] He ran a railroad. The railroad became big enough that one person couldn't keep track of everything that was going on.
A railroad has to have tight coordination, or it won't work. There are many things happening at different locations which depend on things happening at other locations. Some things had to coordinated on a scale of minutes, some on a scale of hours, days, or weeks.
A railroad needed a distributed system which could, in a sense, think.
This was a new problem. Before that, there were few inter-city organizations, and the ones that existed, such as post offices, governments, and shipping lines, were very loosely coupled.
Each location had a boss, who was mostly autonomous. That didn't work for railroads. Unlike ships, the moving trains are totally dependent on a fixed infrastructure.
Different people deal with the rolling stock, the track, the freight, and the passengers.
McCallum figured out how to make that work.
McCallum came up with daily, weekly, and monthly reports, exception reporting, and line and staff management. This made it possible for ordinary people to do something no one person could comprehend in totality. Most organizations still follow his general approach.
That was the point at which the organization became bigger than any one person.
We're not there yet with AI. Coordinating large numbers of AIs is an unsolved problem.
[1] https://www.mckinsey.com/capabilities/people-and-organizatio...