Integrating generative AI into physiotherapy education: students' use, perceptions, and generative AI literacy following curricular adaptations: a repeated cross-sectional programme evaluation.

BACKGROUND: Structured generative artificial intelligence (GAI) training has been proposed as an effective approach to strengthen healthcare students' confidence in navigating AI-related challenges and ethical considerations. However, few studies have evaluated the impact of implemented GAI-related curricular adaptations and learning activities, particularly within physiotherapy education. This study aimed to evaluate: 1) physiotherapy students' use and perceptions of GAI in their studies during implementation of GAI-related curricular adaptations; and 2) GAI literacy in semester 1 before participation and in semester 3 after completion of structured GAI learning activities. METHODS: A two-part repeated cross-sectional programme evaluation with non-equivalent, unlinked samples was conducted among physiotherapy students at Linköping University, Sweden. Students' use and perceptions of GAI were assessed after implementation of GAI-related curricular adaptations in 2024 and again in a separate survey wave in 2025. GAI literacy was evaluated using the 10-item Generative Artificial Intelligence Literacy Scale (GAILS-10) before and after the GAI learning activities delivered during semesters 1-3. RESULTS: Reported GAI use differed significantly between 2024 and 2025 surveys, χ2(3) = 62.40, p < 0.001, with daily use reported by 57% of respondents in the 2025 survey compared with 13% in the 2024 survey. Respondents in the 2025 survey reported more diverse use of GAI, including brainstorming, question generation, and text refinement (p < 0.001). Perceptions of GAI as contributing to learning were also more positive in the 2025 survey, χ2(4) = 24.94, p < 0.001, whereas perceptions of information about permissible and impermissible use were similar in both surveys. Semester 3 students assessed after completing the GAI learning activities reported higher GAILS-10 scores across all dimensions than semester 1 students assessed before participation, with a large between-group effect size for the total score (d = 1.0). CONCLUSION: Physiotherapy students reported more frequent and diverse GAI use and more positive learning perceptions in the 2025 survey than in the 2024 survey. Semester 3 students, assessed after completing the GAI learning activities, reported higher GAI literacy than semester 1 students assessed before participation in these activities. These findings suggest a need for curricular adaptations and educational support to promote appropriate GAI use, reduce misuse, and strengthen student GAI literacy.