NY 8th Grader Zeynep Tests AI Models for Stress Detection

Zeynep Demirbas, a 14-year-old from New York, tested AI models for stress detection. Her study revealed that MentalBERT outperformed ChatGPT-4o, achieving 82% accuracy compared to 74%. These findings raise concerns about the reliability of general-purpose AI models for mental health assessments, highlighting the need for specialized tools in this field.