The Apprenticeship Problem

You need experience to supervise AI.

But AI eliminates the entry-level roles where people gain experience.

In 5 years, who becomes senior?

We automate the grunt work first. The boring stuff.

Data entry. Basic analysis. Junior copywriting. First drafts. Research tasks.

Makes sense. That's what AI does best. That's where the ROI lives. That's what shareholders want.

But here's what we're actually automating: the learning ground.

Junior roles aren't about the output. They're about the reps.

You do the boring work badly. Then less badly. Then competently. Then you see patterns. Then you know why the patterns matter. Then you become the person who can tell AI what good looks like.

We're cutting the bottom rungs off the ladder.

Then wondering why nobody can climb it.

In five years, we'll need people who can prompt AI, evaluate its output, and fix its mistakes. People with judgment. With context. With taste.

Where do those people come from?

The junior roles we eliminated.

We optimized for today's efficiency. We traded tomorrow's capability.

The market will figure this out. Just not in time.

Companies will realize they have senior people retiring and nobody ready to replace them. They'll realize AI needs supervision and nobody knows how to supervise it. They'll realize expertise isn't a prompt. It's ten thousand hours of boring work that taught you what matters.

By then, the gap will be real. The pipeline will be empty.

The institutions that created expertise will have spent a decade not creating it.

We automated away the apprenticeship.

Then we'll wonder why there are no masters.

The system works perfectly.

It just doesn't work for the future.