Sri Lankan medical undergraduates report widespread AI use, with limited formal training

A cross-sectional online survey of medical undergraduates at one Sri Lankan medical faculty found widespread reported use of artificial intelligence applications despite limited formal training. The study analysed 369 responses from Years 1–5 at SUSL in May 2025, assessing AI tool use, self-rated proficiency, attitudes toward AI in medical education and patient care, and support for curriculum integration. The results are reported in BMC Medical Education.

Among the 369 respondents, 366, or 99.2%, reported using AI-based applications. Of these respondents, ChatGPT was the most frequently used tool at 98.1%, followed by Google Gemini at 35.0%, DeepSeek at 20.2% and Microsoft Copilot at 6.3%. The main uses were subject-matter reference, reported by 89.3% of AI users, general information retrieval by 73.0%, assignment assistance by 50.8% and writing improvement by 39.6%. Self-rated proficiency was highest for ChatGPT, with 92.4% of its users rating their proficiency as good or higher. In an analysis of 360 respondents, 77.5% reported using Medscape, 37.2% used UpToDate and 15.3% used WebMD or WebMD Symptom Checker.

Most students agreed that AI could enhance learning in preclinical subjects, with 85.1% expressing agreement, compared with 77.8% for paraclinical subjects and 68.6% for clinical subjects. The questionnaire contained 12 Likert-scale items covering curriculum integration, physician roles and patient care. After adjustment for multiple comparisons, agreement with including AI in the medical curriculum differed significantly across year groups, according to a Kruskal–Wallis test (H = 29.00, Holm-adjusted p < 0.001). The abstract also reports differences in attitudes toward AI for selected items between preclinical and clinical students, with higher agreement among some earlier-year groups for curriculum support.

These findings support consideration of structured, ethically grounded AI education that addresses students’ needs for safe, effective and responsible use of AI. The study is limited by its cross-sectional design, self-reported online responses and focus on undergraduates at a single Sri Lankan medical faculty. It describes reported use, proficiency and attitudes at one point in time; it cannot establish longitudinal change, causal effects or the effectiveness of any educational intervention. The authors’ conclusion is therefore a consideration for curriculum development rather than evidence that a particular training programme improves outcomes.

Subject of Research: Medical students’ use and perceptions of artificial intelligence

Article Title: Artificial intelligence use, perceptions, and educational needs among medical undergraduates at a Sri Lankan medical faculty: a cross-sectional survey

Article References: Andadola, A. M. S. U., Walallawita, W. K. M. M. M., Perera, P. D. V. M., & Mendis, S. N. J. K. (2026). Artificial intelligence use, perceptions, and educational needs among medical undergraduates at a Sri Lankan medical faculty: a cross-sectional survey. BMC Medical Education. https://doi.org/10.1186/s12909-026-10496-0

Image Credits: AI Generated

DOI: 10.1186/s12909-026-10496-0

Keywords: artificial intelligence, medical education, medical undergraduates, Sri Lanka, AI literacy, student perceptions, curriculum integration, ChatGPT, health information technology, Artificial, intelligence, perceptions