Students Using AI Tools Score Lower Than Peers — But Usage Pattern Matters
The PISA 2025 Findings
The Organisation for Economic Co-operation and Development (OECD) has released its first PISA (Programme for International Student Assessment) data since artificial intelligence became widely available to students. The study, based on 2025 data collected from 15-year-olds across numerous countries, reveals an unexpected pattern: students who use AI to help with their studies tend to perform worse in science, math, and reading assessments than those who don't.
Why the Counterintuitive Result?
Researchers caution against oversimplifying the headline finding. The relationship between AI use and academic performance is more complex than it initially appears. Simply delegating homework or study tasks to AI tools without engagement appears to diminish learning outcomes rather than enhance them.
A Nuance Worth Noting
However, the data isn't uniformly negative. Certain approaches to AI use show a different pattern. Students who were explicitly taught to critically evaluate AI-generated responses—to assess accuracy, identify gaps, and verify information—demonstrated slight improvements in their scores compared to similar students who didn't use AI at all.
This suggests that AI's impact on learning may depend heavily on how the tools are integrated into educational practices. Passive consumption of AI outputs appears to underperform, while active, critical engagement with AI assistance shows potential for modest benefit.
Implications for Educators
The findings point to a need for structured guidance on AI use in schools. Simply providing access to AI tools without teaching students how to critically interact with them may not yield positive educational outcomes. The PISA data implies that AI literacy—understanding both capabilities and limitations—could be as important as the technology itself.
Further analysis of the dataset is expected, with researchers hoping to isolate which specific usage patterns lead to better or worse outcomes across different subjects and student demographics.