Inteligencia artificial como entrenador virtual: efectividad comparativa del feedback automatizado versus métodos tradicionales en Educación Física

Introduction: Physical education is essential for an holistic child development, fostering physical, cognitive, and emotional growth. As technology reshapes education, the potential of AI to enhance this vital subject is increasingly recognized. The advent of language models like GPT-3 has opened opportunities for innovative and interactive learning experiences in physical education. While these models demonstrate remarkable capabilities in generating human-like text, their potential as virtual coaches remains largely unexplored. Objective: To assess whether AI can provide athletes with personalized feedback comparable to human coaching. Methodology: The study employed Walter+, an AI tool designed for educational settings, in a pilot trial with three groups: professor-led (TG1), notes-based (NG2), and chatbot-assisted (WG3). Motivational, autonomy-related, and academic performance variables were analyzed. Results: Interactive learning (TG1 and WG3) outperformed passive note-taking (NG2) in motivation and engagement. WG3 excelled in autonomy support (32.00) and autonomous motivation (21.00), while TG1 led in intrinsic motivation (24.00) and competence (27.67). NG2 showed marginal academic gains (40% exam score) but scored lowest in psychological and behavioral metrics. Discussion: Chatbots proved highly effective in fostering autonomy and intrinsic motivation, nearly matching professor-led interaction. Passive learning underperformed in engagement despite slight test advantages, highlighting the limitations of rote methods. The findings position AI as a promising tool for interactive learning. Conclusions: The results indicate that interactive learning, whether with a professor or a chatbot, is more effective than passive learning using only notes. Each method has its own advantages and may be more suitable depending on the specific educational goal.