Ninety days ago, the story was "AI is going to replace everyone."

Now the story is quieter. It's the companies that over-cut headcount, sitting with the early regret.

Both stories miss the same thing.

Entry-level work isn't just cheap work. It's where managers get made. It's where a person learns how the job actually works, so that ten years later they can run it, teach it, and catch it when it breaks.

Cut all of that out and hand it to AI, and here's what you've actually done. You saved money this year and deleted your leadership pipeline for the next decade. You can't manage work you've never done. You can't train a person, or a machine, to do a job you don't understand yourself.

The receipts already back this up. MIT studied 300 enterprise AI deployments in 2025 and found 95% of the pilots delivered no measurable impact on the P&L. Somebody still has to build the rails, feed the data, and catch the thing when it's confidently wrong. That somebody is a person who learned the fundamentals the slow way.

And when the real test comes, an ice storm, a plant down, a system failure at 2 a.m.- you don't need a model. You need a human who has done the work and knows what "right" looks like.

A lot of what's getting blamed on AI right now is just the COVID over-hiring correction wearing a costume. Cutting is easy. Cutting capability by accident is the expensive part.

So before the next round of "AI can do that," run the harder math. Pulling out every seat where people learn the business saves you today and leaves no one to run it in five years.

Where are you cutting costs, and where are you quietly cutting capability?