CONCEPT
The Apprenticeship Problem in AI-Era Software
The structural concern that AI-generated code eliminates exactly the implementation work through which diagnostic capability was traditionally built — leaving a rising generation of developers whose daily tasks do not produce the geological strata that their eventual encounter with a leak will demand.
The apprenticeship problem is the specific software-engineering instance of a broader pattern already articulated in the general concept entry: professional expertise historically formed through sustained engagement with the friction of the practice, and tools that eliminate the friction also eliminate the apprenticeship through which the next generation's expertise was built. In software, the friction was the implementation layer — writing code, debugging code, reading other people's code, understanding why systems worked and why they failed. AI-generated code eliminates that layer from daily practice. The junior developer who enters the profession in the AI era does not go through the implementation-layer apprenticeship that produced her predecessors' diagnostic intuition. She may be extraordinarily productive at the abstraction level. She has no path to the capability that leaks will demand of her.
In The You On AI Field Guide
The apprenticeship model in software engineering was