According to 2025 data from the Programme for International Student Assessment (PISA) conducted by the Organisation for Economic Co-operation and Development (OECD), students who use AI for learning tend to have lower school performance compared to those who do not. This study, targeting 15-year-olds across 91 countries, is the first to be conducted after the widespread adoption of AI.
NewsOrganisation for Economic Co-operation and Development
AI Use Tends to Lead to Lower Academic Performance, OECD's PISA Study Reveals
This article is a translation. Read the Japanese original
Even after adjusting for socioeconomic backgrounds, students who do not use AI at all tended to achieve higher scores in areas such as science. In particular, students who used AI for specific tasks, such as summarizing reading materials or drafting reports, showed relatively lower performance. On the other hand, students who used AI for broader purposes, such as preliminary research or "to help with learning," experienced a smaller decline in performance.
The frequency of AI use also had an impact: those who used AI only once or twice a year, as well as those who used it daily, tended to have lower scores, while those who used it once a month or once a week showed better results. However, in cases where students use AI intentionally—such as those trained to evaluate the quality of AI-generated information—they may achieve higher scores than students who do not use AI. Andreas Schleicher, Director for Education and Skills at the OECD, stated that if technology creates a shortcut rather than supporting the thinking process in learning, it may hinder educational development.
Sources
- Students who use AI generally score worse at school （The Verge AI, 2026-09-09）