Measuring AI preparedness in health professions education: Evidence from a national survey of medical radiation science students and new graduates

Introduction
Artificial intelligence (AI) in medical radiation science (MRS) is increasingly embedded in everyday clinical workflows. As AI systems assume more operational roles, questions arise not only about technical competence, but about professional judgement, ethical responsibility, and what it now means to be “ready for practice”. This study benchmarks perceived AI preparedness among Australian MRS students and new graduates.
Methods
A national cross-sectional online survey was conducted with final-year students and recent graduates from accredited Australian MRS programs. The Medical Artificial Intelligence Readiness Scale for Medical Students (MAIRS-MS) assessed preparedness across Cognition, Ability, Vision, and Ethics. Internal consistency and factor structure were evaluated, and group differences examined by exposure to formal AI education and AI tools during placement. One open-ended question explored participants’ reflections on preparedness and education, analysed using reflexive thematic analysis.
Results
Seventy-eight participants responded (72 with complete MAIRS-MS data). The MAIRS-MS demonstrated acceptable-to-excellent internal consistency and a well-fitting four-domain structure. Ethics scored highest and Cognition lowest, with Ability and Vision intermediate. Participants who reported receiving formal AI education had higher preparedness scores than those who did not (79.1 ± 12.0 vs 70.5 ± 13.0, p = 0.01), as did those with clinical exposure to AI tools during placement (80.1 ± 11.2 vs 69.5 ± 13.0, p < 0.01). Qualitative analysis identified four interrelated themes: variable confidence and readiness; ethical responsibility and professional identity; navigating different “styles” of AI in education and practice (clinical AI vs generative AI); and structural misalignment between university teaching and clinical realities.
Conclusion
Graduates expressed strong ethical orientation toward AI use but weaker confidence in foundational AI knowledge and applied understanding. This imbalance may limit critical appraisal of AI outputs in clinical practice.