The expert forgets their own profession
„The doctor, developer, banker and leader who leave judgment to the model may gradually lose the unassisted expertise needed to catch a faulty machine suggestion. For doctors, one review already reports a measurable decline within three months.”
Counter-narrative
Skills are being reorganized, not disappearing: for young programmers, problem-solving thinking is taking precedence over coding. Pilots' experience with autopilot offers lessons for managing skill loss. This view is represented by training specialists, tech educators, and part of the programming community. One analysis suggests adaptation metrics do not show a decline in judgment.
Carriers
American College of Physicians and medical journals, Goldman Sachs (Chris Churchman, head of the Marquee platform), IBM, BCG and HR consultants, Anthropic Claude Code csapata (Addy Osmani), Trade press around the UK financial regulator (FCA), Financial Times and HR trade press
Analysis
This is a narrative of professions' self-definition. The essence of expertise is no longer possession of knowledge, but the ability to control the machine, and this ability may erode precisely through machine use. It is fueled by clinical measurements and statements from critical professions (medicine, finance, regulation). Professional bodies have an interest because it returns the gatekeeper role to them: a study suggests regulation involving professional communities. It opens a training market for consultants. Its weak point is that most workplace data measures fear, not performance, and there is little comparison with non-AI-using controls.
What's at stake: Whether the layer of experts capable of spotting machine errors will remain.
Narrative sub-threads
Doctors' declining independent clinical judgement
Reports indicate declining performance of AI-using doctors without assistance. One summary states their independent skills dropped by 6 points after three months of AI use. Another report says AI-trained doctors' clinical decision accuracy decreased when given incorrect, LLM-generated suggestions.
Here, performance data indicating decline is also noted, though known only from secondary summaries. A professional body, the ACP, identified the prevention of skill loss in a statement.
→ lasting · evidence: weakDevelopers who do not understand their code
The hidden risk of rapid AI code generation is that teams adopt solutions whose foundations they do not understand. According to Addy Osmani, anyone who cannot explain the answer even with the help of an agent has relinquished ownership. Meanwhile, for young programmers, the required knowledge is shifting from coding to problem-solving thinking.
It narrows to software development and junior skill profiles; evidence is practitioner statements, not measurement.
→ lasting · evidence: weakThe worn-out arguments of executives and bankers
Warnings say managers' and bankers' judgment and reasoning may erode if they delegate analysis and decisions to models. An SBS–MIT study indicates a decline in critical thinking among young salespeople. One analysis, however, says adaptation metrics do not show a decline in judgment.
It concerns high-stakes decision-makers. Financial players also raise the risk: Chris Churchman (Goldman Sachs) spoke to CNBC, and the risk of 'cognitive surrender' arose in connection with the UK FCA becoming an agentic regulator.
→ lasting · evidence: moderateEmployee fear of skill loss and organizational response
In surveys, workers fear that AI will degrade their skills. According to IBM, 60% of employees fear this, most often citing critical thinking. Solutions proposed include skills mapping, clear AI boundaries, pilot experience, and regulation. Meanwhile, many companies do not integrate critical thinking into their training, and nearly half of organizations do not involve HR leaders in AI decisions.
The actors are masses of knowledge workers and HR; evidence is mainly self-reported surveys, not performance metrics.
↗ rising · evidence: moderateThe expert takes over the machine's argument
Research suggests that alongside AI explanations and agents, people tend to outsource oversight as well. Experts may unknowingly adopt machine arguments, even treating them as their own. An arXiv study found that mandatory written reasoning reduced students' uncritical acceptance of flawed AI suggestions without harming acceptance of good ones.
It is not about a profession, but about the mechanism of flawed proposals slipping through, with research evidence and an antidote tested on students.
Sub-thread stories
Doctors' declining independent clinical judgement
Developers who do not understand their code
The worn-out arguments of executives and bankers
Employee fear of skill loss and organizational response
The expert takes over the machine's argument
Google search interest
Hungary
| Topic and search term | Weekly | Average |
|---|---|---|
| The expert forgets their own professionAI-driven skill loss | 0 | |
| Doctors' declining independent clinical judgementartificial intelligence medicine | 0 | |
| Developers who do not understand their codevibe coding | 5.5 | |
| The worn-out arguments of executives and bankersAI decision-making | 0 | |
| Employee fear of skill loss and organizational responseartificial intelligence workplace | 0 | |
| The expert takes over the machine's argumentautomation bias | 0 |
World
| Topic and search term | Weekly | Average |
|---|---|---|
| The expert forgets their own professionai deskilling | 0 | |
| Doctors' declining independent clinical judgementai doctors deskilling | 0 | |
| Developers who do not understand their codevibe coding | 48.4 | |
| The worn-out arguments of executives and bankersai decision making | 10.9 | |
| Employee fear of skill loss and organizational responseai skills fear workers | 0 | |
| The expert takes over the machine's argumentautomation bias | 3.2 |
Google Trends relative search index over the period studied, weekly average. A narrative and its sub-threads appear in a single comparison: 100 is the group's most-searched day, and the figures can only be compared within the group. The search term was chosen by machine analysis; it measures how much the term is searched, not how widespread the narrative is.
Hungary
Domestic presence during the period under review was minimal, with a single event. At the Portfolio HR Summit, Eszter Kovács, who researches cognitive abilities in collaboration with foreign universities, stated that AI does not make us useless, but it matters how we use it. The report suggests that debate should be entrusted to AI instead of thinking. It is worth monitoring whether domestic medical professional bodies and financial regulators will adopt an ACP- or FCA-type approach.
Weekly trend
Turning points
Related narratives in the News magazine
Supporting stories
Machine analysis based on story titles and summaries, with a verification pass. The source of the facts is the article linked at each story.