A junior engineer hits a wall. The database migration script is failing with an obscure error. Five years ago, her manager would have pulled over a senior developer, who would have talked her through the ORM’s quirks while they debugged it together. An hour later, the script would be fixed, and more importantly, the junior engineer would have absorbed a piece of critical, unwritten knowledge about the system.
Today, she is told to paste the error into the company’s AI coding assistant.
It spits out a confident block of code. She uses it. The script runs without error, but it introduces a subtle data corruption bug that a unit test won’t catch for weeks. The junior engineer learned nothing about the database. The senior developer was never interrupted; he was busy answering emails. Productivity, on paper, went up.
This is how expertise dies. The path from novice to master is not a series of lectures but a thousand solved problems. We are aggressively automating the learning process out of existence. The middle-rung, journeyman tasks—the tedious debugging, the boilerplate generation, the first-draft report writing—are the very exercises that build professional intuition. They are the reps. By handing them to a machine, we are skipping the workout and wondering why we’re not getting stronger.
The problem is that AI is best at the tasks of a competent amateur. It can write a passable Python script, summarize a document, or draft a marketing email. These are precisely the jobs we give to people on their way to becoming experts. A law firm associate learns contract law by reviewing hundreds of routine agreements. A marketing coordinator learns strategy by writing dozens of press releases. A
Generated by Reportify AI — Automate your team's status reports, standups, and weekly updates. Try free →