Everyone is counting what AI gives them. Few are counting what they're trading away.

I'm not talking about jobs. The jobs argument is real but it's also the visible part. What's harder to see is what disappears when you remove the work that builds expertise in the first place.

Watch what's happening with manufacturing reshoring right now. Leaders want to bring production back and are discovering you can't just decide to rebuild what was offshored. The expertise didn't get archived. It dissolved. The supply chains that supported it dissolved. Rebuilding is a generation-long project.

We are making the same decision about cognitive labor right now. At greater scale. And faster.

When AI handles the work that junior analysts, lawyers, financial analysts, and engineers learn from, we aren't freeing them up for higher value work. We're removing the gym where the mental muscles for that higher value work get built. The friction isn't the problem. The friction is the education.

Nobody in the system is acting irrationally. Executives reduce costs. Tools get deployed. Junior staff learn to work with AI conclusions rather than develop their own. The expertise pipeline quietly collapses anyway — as an emergent property of individually reasonable decisions.

This isn't an anti-AI argument. Speed and scale matter, and organizations that ignore these tools will fall behind. This is an argument about irreversibility. You can decide to automate the developmental pathway today. Recreating it when you need people who can think independently, and discovering you don't have them, is a different problem entirely.

The gym closes permanently once the weights become weightless.