That seems like one of the more reasonable scenarios to expect “AI” to thrive, actually.
You have dozens to thousands of people doing the actual work, then you have one CEO or owner that asks the AI to summarize what all the workers have been doing and communicating. It then presents them with a list of suggested decisions with some metrics tied to them like chance of success and Net Present Value (NPV).
The AI wouldn’t even have to do a good job! It would just have to get things wrong or make up bullshit whole cloth as much as human middle managers, or maybe even less!
I think this is a possiblity not because the AI will do a good job, but because the ONE thing that LLMs seem to be good at is mimicking what a random human might type, including humans who are often ignorant or wrong.
What happens when middle management realizes the only jobs AI can actually do is their jobs?
That seems like one of the more reasonable scenarios to expect “AI” to thrive, actually.
You have dozens to thousands of people doing the actual work, then you have one CEO or owner that asks the AI to summarize what all the workers have been doing and communicating. It then presents them with a list of suggested decisions with some metrics tied to them like chance of success and Net Present Value (NPV).
The AI wouldn’t even have to do a good job! It would just have to get things wrong or make up bullshit whole cloth as much as human middle managers, or maybe even less!
I think this is a possiblity not because the AI will do a good job, but because the ONE thing that LLMs seem to be good at is mimicking what a random human might type, including humans who are often ignorant or wrong.