MoveBench evaluates large language models for safe sports coaching
MoveBench: A Rubric-Based Benchmark for Evaluating Large Language Models in Sports and Health Coaching
Artificial Intelligence
Summary
Large language models (LLMs) like AI chatbots can help with sports and health advice, but they may not always give safe or personalized recommendations. The authors created MoveBench, a special test that checks how well these models perform in real coaching scenarios, focusing on safety and scientific accuracy. They tested 15 different models and found big differences in how well they handle important coaching tasks. This work helps build better, safer AI tools for fitness and health coaching.
What this means in practice
- •For fitness app developers: Evaluate language models to ensure safe and personalized exercise recommendations in fitness apps using a domain-specific benchmark.
- •For health technology companies: Test AI coaching systems for compliance with scientific standards and user safety before deployment in consumer health products.
Authors
Yitong Li, Jiang Bian, Jingling Wang, Pan Ding, Chengying Gu
Abstract
Despite rapid advances in large language models (LLMs), their deployment in the sports and health domain remains high-risk due to the absence of domain-specific evaluation benchmarks. Existing general-purpose benchmarks (e.g., MMLU and HELM) primarily assess broad knowledge and reasoning capabilities, but fail to capture critical domain requirements such as exercise safety, physiological appropriateness, personalization, and actionable coaching quality. This gap is particularly concerning in fitness and health applications, where incorrect or unsafe recommendations can lead to physical harm. To address this limitation, we introduce MoveBench, the first comprehensive benchmarking framework specifically designed for evaluating LLMs in sports and health coaching. The proposed benchmark integrates heterogeneous task formats, including multiple-choice questions (MCQs) and open-ended question answering (OpenQA), grounded in real-world user scenarios and paired with structured rubrics emphasizing safety compliance, scientific validity, and training logic. Through large-scale evaluation of 15 diverse LLMs on 809 questions with 5,012 rubric-level criteria, our results reveal substantial performance variability, particularly in safety awareness, structured program design, and personalization capability, highlighting a critical gap between knowing and doing. Our work establishes a foundational evaluation standard for sports and health LLM applications and provides a reproducible framework for advancing safe, reliable, and context-aware AI-driven coaching systems. Open resources are available at https://github.com/SatoSakula/MoveBench.