The Generative AI Learning Penalty: Evidence from Chinese Secondary Education

Abstract

Using 30 months of panel data on 26,811 Chinese students in grades 7-12, we study how generative AI a¤ects homework productivity and learning. The data combine monthly closed-book exams, high-school and college entrance exams, and homework scores and completion time across nine subjects. We exploit staggered AI adoption in a di¤erence-in-di¤erences design. AI adoption raises homework scores by 18% and reduces completion time by 30%, but lowers monthly exam scores by 20% within six months. High-stakes entrance-exam scores fall by 18 and 24%, with the full penalty emerging only after about two years. The losses are largest in social science subjects, followed by STEM and languages, and are especially large for junior students, high-achieving students, and boys. The learning losses are concentrated among roughly 80% of AI users whose behavior is consistent with homework outsourcing, as indicated by exceptionally short homework completion time coupled with high homework scores. AI users who maintain similar homework completion time as non-AI users experience small learning losses. The authors have no sources of research support, …nancial relationships, or other potential con ‡icts of interest to disclose. The data collection and usage of this project were approved by the Human Research Ethics Committee of the University of Hong Kong with the following reference number: