The craftsmanship question is real, but I think the more pressing issue is what replaces it. Craft wasn't just about aesthetics — it was about understanding the system deeply enough to make good architectural decisions, trace bugs, and reason about failure modes.

When that depth erodes, you don't just lose "nice" code. You lose the ability to maintain it, extend it, or even know when something has gone wrong.

A few of us have been thinking about this and put together the Agile Vibe Coding Manifesto (https://agilevibecoding.org) — an attempt to extend the original Agile Manifesto for environments where AI is doing significant amounts of the generating. One of its core principles: "Generated systems remain understandable and maintainable by humans, regardless of how they were produced."

The manifesto isn't a rejection of AI tools — it's an argument that the disciplines of craftsmanship (traceability, domain clarity, human accountability) matter even more in an AI-assisted world, not less. Worth a read if this resonates.

it's kind of the difference between a studio musician and a songwriter. one you're doing for a paycheck, and you may or may not enjoy the work to varying degrees. the other is your passion.

LLMs are freeing companies to produce faster, but your expertise is still needed to supervise them for now.

coding in your spare time as a passion project is likely the outcome for many engineers who like the craft.

you might go as far as separating environments, or even whole machines, for your passion. keep a machine with NO ai options available to push you to code solutions by hand, etc.