It may not be all doom and gloom for AI in the classroom—there actually may be a Goldilocks effect.

The OECD’s latest Program for International Student Assessment (PISA) found that AI has led to a decline in multiple student skills as well as a drop in test scores, but moderate use is associated with stronger performance in certain subjects compared to limited or daily use.

For several common assignments, students who used AI “moderately”—from about once a month to once or twice a week—recorded higher average science scores than both students who rarely used AI and those who used it almost every day. The advantage was especially clear when students used AI to summarize assigned reading or conduct preliminary research on a new topic.

Students who reported using AI for that purpose weekly tended to perform at about the same level as AI nonusers after accounting for socioeconomic status. But students who used it almost every day and students who rarely used it, tended to have lower scores.

“These relationships do not necessarily imply a negative impact of AI use on science performance, but may reflect a complex mix of who adopts AI and how they use it,” the study found. “It also suggests that moderate and intentional use of AI for schoolwork and learning may be positively related to student performance.”

The OECD says its findings are consistent with the idea that digital tools can support learning when used purposefully and in moderation, but excessive or unguided use can contribute to distraction and shallower engagement.

“In the same way that we do not become fit by watching sports but by doing sports, learning does not occur through the consumption of content,” the findings read, “but as a productive cognitive struggle of the mind with new material.”

But lower scores are not all due to AI. PISA has reported that reading, mathematics and science scores have dropped since 2003.

Still, the latest results showed that 86% of students across OECD countries had used AI chatbots for at least one of the schoolwork purposes examined in the assessment, with the remaining 14% saying they had never or almost never used AI for any of those purposes.

And students who do not use AI chatbots for schoolwork generally score higher than students who do, according to PISA.

Meanwhile, about one in five students reported using it almost everyday. This group tended to score lower in science than students who did not, accounting for socioeconomic status.

That creates a feedback loop for students who rely too heavily on AI. The weaker a student’s ability to read critically, the harder it may be to recognize when an AI-generated answer is incomplete or incorrect. And at the same time, routinely allowing AI to perform the reading, summarizing or reasoning could give students fewer opportunities to develop those skills.

And the findings also say important skills attributed to critical thinking are declining.

“Most troubling is the decline in the very skills that matter most in the AI age: evaluating information, making connections across multiple sources, and thinking critically about what we read,” the report read. “In turn, the share of ‘hasty readers’—students who rush through texts and give quick but incorrect answers—almost doubled between 2018 and 2025. In a world increasingly flooded with information, our students are less likely to think deeply and separate signal from noise.”