Evaluating Large Language Models as Mental Health Support Tools for Teenage Athletes

This study investigates the potential of Large Language Models (LLMs) as mental health support tools for teenage athletes aged 13-18. Through a survey of 60 teenage athletes across 15 sports, we identified four primary mental health challenges: 68% mentioned performance anxiety, 80.2% mentioned peer pressure, 48% of them had confidence-related stress, and 39% had social comparison/perfectionism issues. Our findings reveal a significant gap between mental health needs and available support, with 78% of athletes attempting to "push through" challenges independently despite 47% expressing need for expert guidance. We evaluated three leading LLMs (ChatGPT-4, Claude 3, and Gemini Pro) using 20 scenario-based prompts derived from survey responses. Results indicate that LLMs show promise in providing accessible, 24/7 support that meets athletes' needs for help during non-traditional hours, with 64% of surveyed athletes open to AI-based support and 73% preferring a hybrid AI-human model. This research contributes to understanding how technology can bridge gaps in mental health support for young athletes while highlighting necessary safeguards and limitations.