Research HighlightPublished: 02 September 2026AI diagnosticsFrank Sun1 Nature Sensors volume 1, page 759 (2026) Cite this articleSave articleView saved researchThe first study involved 623 members of the general public performing a binary classification task to distinguish melanoma from nevus, whereas the second involved 153 primary care physicians completing a differential diagnosis task across four skin conditions. An additional group of 320 medical students completed the same task as the physicians to assess the effect of medical training. In both studies, participants were randomly assigned to one of four AI assistance methods.Among the general public, AI assistance improved diagnostic accuracy from 69.7% to 75.8% and reduced performance differences across skin tones. Multimodal large language model (LLM) explanations provided the greatest improvement when AI predictions were correct, but also increased trust in incorrect AI predictions, leaving participants more likely to become overconfident in wrong diagnoses. Among primary care physicians, AI assistance substantially improved diagnostic performance. Unlike the general public, however, physicians were largely resistant to incorrect AI recommendations, relying on their own clinical judgment even when the AI was wrong, and showed greater resilience than medical students. Clinical expertise helped physicians use AI more critically, allowing them to benefit from AI support without being easily misled by incorrect explanations.This is a preview of subscription content, access via your institutionAccess options Access through your institutionSubscribe to this journalReceive 12 digital issues and online access to articles118,99 € per yearonly 9,92 € per issueLearn moreBuy this articlePurchase on SpringerLinkInstant access to the full article PDF.39,95 €Prices may be subject to local taxes which are calculated during checkoutSubjectsMachine learningPredictive medicineSkin manifestationsAuthor informationAuthors and AffiliationsNature Sensors https://www.nature.com/natsensors/Frank SunAuthorsFrank SunView author publicationsSearch author on:PubMed Google ScholarCorresponding authorCorrespondence to Frank Sun.Rights and permissionsReprints and permissionsAbout this article