AI-assisted learning in exercise physiology: a quasi-experimental study using PhysioExercise GPT

This study evaluated the applicability and effectiveness of PhysioExercise GPT, an artificial intelligence tool based on the ChatGPT architecture, developed to support the teaching of exercise physiology to undergraduate physical education students. A longitudinal quasi-experimental study was conducted involving a total of 64 students. Thirty-two students used the AI tool through a structured educational approach that incorporated personalized support, immediate feedback, and learning activities based on Bloom's Taxonomy, while a control group of 32 students received traditional instruction. The results showed that both groups improved over time; however, the AI-assisted group achieved significantly greater learning gains (p = 0.002). The intervention group demonstrated a mean improvement of 3.38 points compared with 1.97 points in the control group, corresponding to a large effect size (Hedges' g = 0.80). Analysis of covariance (ANCOVA) confirmed that the superiority of the AI-assisted group remained significant after adjusting for baseline performance. The findings indicate that the structured use of PhysioExercise GPT functions as an effective digital tutor, promoting active learning and enhancing the retention of complex concepts. Although the study has limitations, including the absence of randomization, the results suggest that AI represents a robust complementary educational tool capable of improving academic performance in health education.