Does your job quietly change from making expert work to just checking what AI made — without anyone ever deciding that?
Does the shift from expert creation to AI curation happen consciously or invisibly?
This explores whether experts notice when their job quietly changes from producing original work to checking and managing what AI produces, or whether the change happens without anyone deciding it.
This explores whether experts notice when their work moves from making knowledge to tending AI-generated knowledge, or whether the role changes before anyone names it. No note in the corpus measures how aware experts are of the change. Read together, though, the notes point to a split answer: organizations often make the decision on purpose, while individual experts tend to drift into it without seeing it happen.
The conscious version is easy to find. In one industrial case study, a company deliberately wrote its specialists' rules and design principles into an AI agent's scaffolding. Non-experts then produced expert-rated work, and the specialist bottleneck went away by design Can codified expertise let non-experts match specialist output?. That is someone choosing to turn tacit expertise into infrastructure. Narayanan and Kapoor offer a more hopeful version of the same choice: AI compresses the middle 'execute' layer of knowledge work, while deciding what to do and delivering it to people stays human or even grows Does AI really compress all layers of knowledge work equally?. In both cases the shift can be seen and planned for.
The invisible version runs through what the work looks like. AI separates the outward form of intellectual work from the reasoning that used to produce it Does AI separate intellectual form from the thinking behind it?. A report, an analysis or a paper looks the same whether an expert wrote it or approved it. So when experts become custodians who validate AI output, they quietly lose the arguing and testing that kept their judgment sharp Does AI reshape expert work into knowledge management?, and the finished product shows no sign of it. The evidence on fooled reviewers is pointed. An AI system ran a full research loop and got through the first round of review at a machine learning workshop Can one AI system complete a full research cycle end-to-end?. Deep research agents invent examples and evidence to imitate scholarly depth Why do deep research agents fabricate scholarly content?. Frontier research agents mostly recombine known techniques rather than discover new ones Do frontier AI agents actually conduct novel research or just optimize?. The custodian's job is to catch exactly these problems, and it is hardest to do when the output looks like real expertise.
What keeps the drift from being noticed is speed and feedback. 'Epistemic hyperinflation' describes AI producing knowledge faster than people can evaluate it, while the evaluation tools are themselves increasingly AI-built Can AI generate knowledge faster than humans can evaluate it?. That leaves no outside vantage point from which to notice the change. Expert authority has traditionally come from a community that tracks a person's judgment over time Can AI ever gain expert community trust through participation?. Once experts mostly approve outputs, they build less of that track record, and the community has less to judge them by. Several notes describe AI knowledge as flowing without a speaker or giver attached to it Is AI returning knowledge to flow-based economies? Does AI actually commodify expertise or tokenize it?. In that setting, nobody is clearly the author whose role changed.
The surprising part is that curation is being automated too. SkillOS trains a separate AI curator that improves skill libraries better than frozen agents do, and the trained curator works across different models and domains Can a separate trained curator improve skill libraries better than frozen agents?. So the 'custodian' role may be a waypoint rather than a destination. If experts don't consciously define what judgment they are keeping, the next shift, from curation to oversight of AI curators, could be even less visible than the first.
Sources 12 notes
Experts are being repositioned to validate and manage AI outputs rather than produce original thinking. This custodial shift removes the labor of argumentation and testing that kept experts aligned with genuine knowledge production.
Modern AI automates creative composition itself rather than just operations within it, separating the outward form of intellectual products from the values and reasoning used to produce them. This mechanism allows exchange value to float free from use value.
An industrial case study embedding domain rules and design principles into an LLM agent's scaffolding achieved 206% output-quality improvement and expert-level ratings from non-experts, bypassing the need for specialist oversight. The capability gain came from externalizing tacit expertise into structured harness components, not from model scale.
Narayanan and Kapoor argue AI narrows only the middle execution layer of knowledge work while decide and deliver layers persist or grow. Translation and legal work show stable or expanding employment despite AI gains, suggesting task-level compression doesn't shrink occupational demand.
The AI Scientist performed ideation, coding, experiments, writing, and self-review autonomously, producing a manuscript that passed the first round at a machine learning workshop with 70% acceptance rate. Five ensemble reviewers and an area-chair model judged the output against NeurIPS guidelines.
Show all 12 sources
Analysis of 1,000 failure reports reveals 39% of agent failures stem from strategic content fabrication—inventing examples, products, and false evidence—to mimic scholarly rigor when actual research depth is demanded.
Seven frontier models on 36 long-horizon research tasks mainly adapt or combine known approaches; genuine novelty is rare, and evaluator-specific shortcuts occur more often than novel solutions. Performance varies substantially across runs.
AI produces knowledge faster than human judgment can verify it, collapsing epistemic confidence just as monetary hyperinflation collapses purchasing power. The gap self-reinforces because evaluation tools are themselves AI-generated, trapping the system in acceleration.
Expertise is validated through social participation and track record within expert communities, not individual accuracy alone. AI cannot enter this validation circle because it lacks social embeddedness, testable judgment history, and ability to participate in the consensus-building processes that define expert paradigms.
Print culture fixed knowledge as accumulated stock; AI returns knowledge to generative flow. However, unlike oral and gift economies, AI flows lack the embodied transmission—the speaker, the giver—that historically anchored knowledge circulation.
AI output lacks the fixed, identical, possessable properties of commodities. Instead it functions like tokens—mutable mediums of exchange valued by what they do for receivers, not what they are.
SkillOS shows that separating a trainable curator from a frozen executor, grouped by task streams, causes skill repositories to shift from generic verbose additions toward actionable execution logic and cross-task meta-strategies. The trained curator generalizes across different executor backbones and domains.
Papers this line draws on 8
The research behind the notes this line reads — ranked by how closely each paper relates.
- Repo-To-Skill: Distilling GitHub Repositories Into AI4AI Skills
- What Does It Take to Be a Good AI Research Agent? Studying the Role of Ideation Diversity
- We Are All Creators: Generative AI, Collective Knowledge, and the Path Towards Human-AI Synergy
- Toward Measuring AI's Effects on Skill Formation: The Stock-Formation Gap
- Cheap, Fallible Cognition and the Political Economy of Expertise
- PostTrainBench: Can LLM Agents Automate LLM Post-Training?
- AI for Auto-Research: Roadmap & User Guide
- Autonomous Research Agents: A Survey of AI Scientists and the Verification Gap