Never-skilling is the failure to form a foundational skill in the first place, because AI did the cognitive work that would have built it. The term comes from a 2026 Nature Medicine Perspective on medical training. It names a risk that its authors say has not yet been demonstrated.
The answer, in one line
The failure to form a foundational skill during training because AI did the cognitive work that would have built it. The term comes from Ke and colleagues in Nature Medicine on 22 May 2026.
Definition#
Never-skilling: trainees who rely on AI during the early formative years of their education may fail to develop the foundational reasoning that independent practice requires. The term is Ke and colleagues’ (Nature Medicine, 22 May 2026). It differs from deskilling, which is the loss of a skill someone once had, because there is no earlier level of competence to return to.
Three failures with three different starting points#
Ke and fifteen co-authors separate three things that are usually run together. Deskilling is the decline of established competence in clinicians who were already trained. Mis-skilling is what happens when “uncritical acceptance of AI errors leads trainees to internalize flawed clinical knowledge as fact”: the capability is built, and built wrong. Never-skilling is the third: the skill is not formed at all. Graded entry.
The distinction changes the remedy. A person who has lost a skill can be put back in front of the work that built it. A person who never built it has nothing to return to, and the first sign may be a situation the tool cannot handle.
A risk model, and its authors say so#
The paper is a Perspective, not a study. It reports no data of its own. The abstract concedes that “direct evidence from medical training is absent” and rests the concern on learning theory and early signals from outside medicine. It also predicts a state it calls false proficiency, competence that looks real while depending on the AI staying available. That prediction is untested as well.
So never-skilling should be cited for the taxonomy and as the origin of the word. Nobody has yet measured how often it happens, how severe it is, or whether it can be reversed.
The nearest measurements are in other people#
Two graded studies sit next to the idea without testing it. Bastani and colleagues gave students unrestricted AI access for practice; their grades rose while the tool was present, and once it was removed the group scored 17 per cent lower than students who had never had it. Graded entry. That is students losing ground against a baseline, not clinicians in training. Budzyn and colleagues found adenoma detection in unassisted colonoscopy fell from 28.4 to 22.4 per cent after AI exposure, but those endoscopists were already experienced, which makes it deskilling rather than never-skilling. Graded entry.
What is missing is the study that follows trainees who learned alongside AI and tests them without it. Until it exists, both the worry and the reassurance are inference.
Why a newcomer can look senior before the skill exists#
This section is interpretation, not finding. If a junior produces senior-looking work through a tool, the output gives no sign of whether the judgement underneath was built. That is the situation synthetic seniority describes, and never-skilling is the name for the mechanism if the answer is that it was not. The two terms come from different people and different fields. Ke and colleagues are writing about clinical education and make no claim about workplaces; applying the idea to a graduate analyst or a trainee lawyer is an extension that the evidence has not yet tested.
A check before the tool is switched off#
For any trainee programme, ask one question of each task the AI now does: if it were unavailable on Monday, could the person do this unaided at an acceptable standard? Where nobody knows, that task is the place to restore unaided practice first. The juniors page sets out what the evidence says about how that practice works.
Key sources
- Ke, Y. et al. (2026). AI-induced never-skilling in medical education. Nature Medicine, 32(6), 1997-2006, 22 May 2026. Graded entry.
- Bastani, H. et al. (2025). Generative AI Without Guardrails Can Harm Learning. PNAS. Graded entry.
- Budzyn, K. et al. (2025). Endoscopist deskilling risk after exposure to artificial intelligence in colonoscopy. The Lancet Gastroenterology and Hepatology, 10(10), 896-903. Graded entry.
Related SuperSkills research#
Start with what deskilling is, then synthetic seniority, the missing rungs and capability debt. The term is Ke and colleagues’, not a SuperSkills coinage.
About this research#
Written by Rahim Hirji, author of SuperSkills (Kogan Page, 2026), keynote speaker on AI and human capability, and founder of The SuperSkills Intelligence Company.
Explainer · SS-2026-419 · Graded against the published rubric
Hirji, R. (2026). What is never-skilling?. The SuperSkills evidence base, SS-2026-419. https://thesuperskills.com/research/what-is-never-skilling. Last reviewed 12 October 2026.
An evidence review by Rahim Hirji, not peer-reviewed research. For a material claim, cite the underlying study as well; every study here carries its own permanent link.
How citations and IDs work