Why some professionals avoid generative AI at work: in a survey of 201 NHS psychology staff, identity threat went with lower intention to use it
Preprints.org (Pye, Hazell, Smith, Bergin), peer-reviewed version listed in Behavioral Sciences 2026, 16(10), 1827 · posted Jul 28, 2026 · 201 UK NHS psychology professionals, cross-sectional survey
Usefulness predicted intention far more strongly than the identity measures did, and the authors report that adding them changed the explained variance by under half a percentage point. Even so, 52 of the 146 coded responses for non-use named threat to professional identity. The two items sit side by side in the data: a tool can be seen as useful and still be linked to a worry about the user's own professional capabilities and sense of self. Whether that worry came first was not something a single survey could test.
Why it matters to this work. The survey asks whether a worry about what an AI tool means for one's own professional standing travels with how much that person uses it. It reports associations in one occupation at one point in time. The Masks
If a tool took over the part of your work you are best known for, what would you want to still be known for?
Find the mask you wear most → A free reading on the ten Masks.The story
IllustrationThe authors surveyed UK NHS psychology professionals, trainees included, between March and July 2025, and asked about generative AI use. They added two identity measures to a standard technology acceptance model: threats to professional capabilities and self-threat. Both showed negative associations with intention to use generative AI (about -.20 and -.16) and with how often it was used in clinical work (about -.14 and -.11). Perceived usefulness was the strongest predictor of intention in every model (.67 to .69). In open-ended answers, 146 coded responses gave reasons for not using it: ethical concerns came up 88 times, threat to professional identity 52 times and lack of trust 42 times.
The sample was 201 people (82.1% female, 79.6% White or White British, mean age 32.3), recruited through social media, psychology conferences and snowball sampling. Adding the identity measures moved the share of explained variance in intention from 78.3% to 78.7%. The authors state that the design is cross-sectional, so causality cannot be claimed, that self-selection may over-represent early adopters, that the sample was too small to test differences by demographic group, and that no validated scale exists for professional identity threat in psychology, so they adapted scales from medical settings.
More from James
- Using AI at work and fear of losing your job: Gallup reports frequent AI users were more than twice as likely to fear job elimination, with a smaller gap where workers felt respectedThe pairing is the finding a headline about AI anxiety would skip: frequent use went with more fear of job elimination, and the gap was smaller among workers who gave their workplace the highest ratings for respect and for care about wellbeing.
- People accepting wrong AI answers: a Wharton paper reports about 80% followed incorrect chatbot advice when they consulted it, and accuracy fell to about 31%The detail a headline would drop is the confidence figure: chatbot access raised confidence even when the advice was wrong.