Artificial intelligence and ChatGPT literacy among surgical healthcare professionals: knowledge, attitudes, and perceived clinical utility

Abstract

Background – Artificial intelligence (AI) and large language models such as ChatGPT are increasingly applied in surgical education and practice. Successful integration requires healthcare professionals to have adequate literacy, perceived knowledge, and agreement in using these tools, alongside awareness of ethical considerations. Evidence on AI and ChatGPT literacy among surgical trainees and specialists remains limited. Objectives – This study aimed to evaluate AI and ChatGPT literacy, perceived knowledge, agreement in clinical application, attitudes, perceived barriers, and perceived clinical utility among surgical medical students, residents, and specialists. Methods – A cross-sectional survey was conducted among medical students (clinical years), surgical residents, and consultants/specialists in Jordan to assess AI literacy, ChatGPT literacy, attitudes, and barriers toward AI integration in surgical practice. Stratified random sampling with proportional allocation ensured representative participation across professional strata. A structured online questionnaire, including the Meta-AI Literacy Scale, ChatGPT literacy items, an AI exposure subscale, and attitude measures, was used. Subscale and overall scores were calculated, and internal consistency was high across all measures (Cronbach's α 0.846–0.930). Data were analyzed using R, with ANOVA or Kruskal–Wallis tests for group comparisons, chi-square tests for categorical associations, and significance set at p < 0.05. Results – Most participants lacked formal AI training (93.2%), but 80.9% had prior ChatGPT experience. AI exposure and perceived literacy scores were highest among students and residents, while consultants reported lower familiarity. Residents showed greater agreement in clinical AI application (MAIL_Use&Apply: 19.5 ± 5.2 vs. 17.4 ± 5.4; p = 0.014) and AI detection (MAIL_AI_Detect:9.0 ± 2.7 vs. 8.5 ± 2.2; p = 0.033) compared with consultants. Ethical awareness was high across all groups (MAIL_AI_Ethics:10.8 ± 2.9). AI was perceived as most useful for surgical planning, diagnostics, and robotic-assisted surgery, with trainees rating utility relatively higher than consultants (p < 0.01). Barriers included lack of training, cost, ethical concerns, and limited infrastructure. Conclusions – AI and ChatGPT literacy varies across professions, with younger trainees reporting relatively higher exposure, agreement, and perceived utility. Ethical awareness is generally high, and attitudes toward AI adoption are positive. Integrating structured AI education and addressing barriers are critical to safe, effective, and ethical adoption of AI in surgical education and practice.