When experts stop doing the thinking and just check what AI produces, do they stay experts or become editors?

This explores what happens to expertise itself when experts stop making knowledge and start vetting AI's version of it. The corpus's sharpest answer is that they keep the title but lose the practice behind it. Experts are being repositioned as custodians of AI-generated knowledge, and that custodial role removes the labor of argumentation and testing that kept them tied to real knowledge production Does AI reshape expert work into knowledge management?. A related note describes the same change as a skill shift from production to validation, alongside inflationary devaluation of the output Is AI fundamentally changing how value gets produced?. Expertise is built by doing the arguing and testing. Reviewing someone else's answer, especially a fluent one, doesn't exercise the same muscles.

Two things make this shift hard to notice. AI separates the outward form of an intellectual product from the reasoning and values that produced it, so a polished report can exist with no expert thought behind it Does AI separate intellectual form from the thinking behind it?. The seamless output also blurs where the human stops and the AI starts, which leads people to treat AI-assisted results as proof of their own skill Do AI-assisted outputs fool users about their own skills?. A custodian can therefore believe they are still expert while the underlying capability erodes. Their checking job is also getting harder. AI generates knowledge faster than people can evaluate it, and because the evaluation tools are AI-generated too, the gap feeds itself Can AI generate knowledge faster than humans can evaluate it?.

Expertise is also social. Experts are validated through participation and a testable track record inside their community, not through individual accuracy alone Can AI ever gain expert community trust through participation?. Meanwhile, heavy generative AI users shift toward solo documentation work, with communication actions growing much less than productivity-app actions Does generative AI shift knowledge workers away from communication?. The notes don't connect these two claims, but together they suggest a compounding cost. A custodian argues and tests less, and does it less with peers, so they build less of the visible judgment history that gave the title its meaning.

Expertise doesn't simply vanish, though. It can move out of people and into systems. In one industrial case, domain rules and design principles were written into an agent's scaffolding, and non-experts then produced expert-rated output, with a 206% quality gain that came from externalizing tacit knowledge rather than from a bigger model Can codified expertise let non-experts match specialist output?. In that picture the expert's lasting contribution is the rulebook, and the person running the system no longer needs to hold it. That fits the argument that AI output behaves less like a fixed commodity and more like a token, valued by what it does for whoever receives it Does AI actually commodify expertise or tokenize it?. The risk is not that experts become obsolete. They can end up holding a validation role whose authority rests on a practice they no longer do.

The one hopeful data point is that when AI touches only a few of a job's tasks, workers can move to the tasks it leaves alone, which softens job losses Does concentrated AI exposure enable workers to adapt and reallocate?. That protection depends on exposure staying concentrated. A custodial shift that spreads across the whole job doesn't get it.