| DAILY NEWS FOR YOU EDUCATION / PISA | |
| |
Global Student Reading and Math Scores Hit Their Lowest PISA Levels. OECD Flags AI Use and Screen Distraction as Growing Risks. |
| PISA 2025 tested more than 760,000 fifteen-year-olds across 91 countries and economies. Reading is down 28 points across OECD countries since 2015, mathematics is down 22, and one in five students is now a low performer in all three core subjects. |
| PISA 2025 / GLOBAL READOUT | OECD | | | 760K+ STUDENTS TESTED | 91 COUNTRIES & ECONOMIES | 20% LOW IN ALL 3 SUBJECTS | | |
| The Organisation for Economic Co-operation and Development released the 2025 edition of the Programme for International Student Assessment this week, documenting another broad deterioration in student performance. PISA 2025 recorded the lowest OECD-average performance observed so far in science, reading and mathematics. Reading and mathematics continued to fall sharply, while average science performance was broadly stable between 2022 and 2025 after a longer period of decline. |
| One in five 15-year-olds across OECD countries is now classified as a low performer in science, mathematics and reading, up from 16 percent in 2022. Reading recorded the steepest long-term decline: average scores fell 28 points between 2015 and 2025, equivalent to roughly one and a half years of learning. Mathematics fell 22 points over the same period, equivalent to just over one year. |
| 2015 → 2025 / OECD AVERAGE | READING −28 score points | MATHEMATICS −22 score points | | |
| PISA 2025 also provides the first major global benchmark of how students are using artificial intelligence for schoolwork. Students reported using AI chatbots to summarize assigned reading, conduct preliminary research, draft written assignments and, more generally, help them learn. |
| AI USE / PERFORMANCE SIGNAL | | ~20 POINT SCIENCE GAP | Students who did not use AI for specific tasks such as drafting, summarizing and research scored about 20 points higher, on average, than students who did. | ASSOCIATION ≠ CAUSATION — OECD says the relationship varies by purpose, frequency and how students are taught to use AI. | |
| For specific schoolwork tasks such as summarizing texts, drafting assignments and conducting preliminary research, students who did not use AI tended to outperform those who did by around 20 points in science — approximately the performance difference associated with one year of schooling. But the OECD cautions against reading this as proof that AI itself causes weaker performance. Students using AI moderately for general learning purposes often performed similarly to non-users, and students who had been taught how to assess AI-generated information sometimes performed better than frequent users without that guidance. |
| |
| Digital distraction is the second major warning running through the report. Across OECD countries, students reported spending about 3.4 hours per day on digital devices for leisure and close to another three hours on learning-related activity during a typical weekday. Moderate educational use was associated with stronger performance up to a point, while very high use and leisure use during lessons were associated with weaker results. |
| 3.4H DAILY LEISURE | + | ~3H DAILY LEARNING | 28% REPORT CLASS DISTRACTION | |
| READING BEHAVIOR NEARLY 2× | The share of “hasty readers” — students moving quickly through texts while producing incorrect answers — has nearly doubled since 2018. | |
| | OECD / MATHIAS CORMANN “Technology and AI can strengthen learning ... but only when used purposefully and not as a substitute for attention, effort and understanding.” | |
| The socioeconomic pattern is also unusual. The gap between advantaged and disadvantaged students narrowed between 2022 and 2025, but largely because higher-income students lost ground rather than because disadvantaged students improved. In reading, advantaged students across OECD countries declined by about 20 points over those three years, compared with a four-point decline among disadvantaged students. |
| READING / 2022 → 2025 | | ADVANTAGED STUDENTS −20 points in reading | DISADVANTAGED STUDENTS −4 points in reading | |
| At the same time, resource pressure remains concentrated in less advantaged schools. Across OECD countries, 39 percent of students attend schools whose principals report that instruction is hindered, at least to some extent, by a lack of teaching staff. The shortage tends to be more acute in disadvantaged schools, making access to sustained human instruction another part of the performance picture. |
| 39% OECD STUDENTS | TEACHER CAPACITY Attend schools where principals say insufficient teaching staff hinders instruction to some extent. | |
| |
| The U.S. results follow the same reading pattern. American 15-year-olds' reading performance fell 14 points from 2022, while mathematics and science were statistically about the same as in the previous PISA cycle. Reading and mathematics are now among the lowest results recorded for the United States since it began participating in PISA. |
| −14 READING VS. 2022 | 74% READING LEVEL 2+ | 65% MATH LEVEL 2+ | |
| DATA NOTE OECD cautions that U.S. participation fell short of several PISA sampling standards in 2025, so the country's results carry greater uncertainty than usual. | |
| 03 | CORE NOTES Three signals inside PISA 2025 | |
| 01 | Students who used AI for specific academic tasks such as drafting, summarizing and research scored around 20 points lower in science than non-users on average. OECD describes this as an association, not proof that AI caused the performance gap. | |
| 02 | The decline among socioeconomically advantaged students was unusually steep. In reading, their OECD-average score fell about 20 points from 2022, compared with four points among disadvantaged students, narrowing the gap primarily because the higher-performing group lost ground. | |
| 03 | PISA's digital findings point less toward abandoning technology than toward how it is used: moderate learning-focused use can coexist with stronger outcomes, while distraction, excessive use and substituting AI-generated answers for sustained cognitive effort are associated with weaker performance. | |
| |
| SOURCES PISA 2025: Students' Reading and Mathematics Performance Declined Sharply Across the OECD — OECD
PISA 2025 Results, Volume I: Future-Ready Students — OECD
PISA 2025 Results: United States — OECD
Teen Reading Slumps to Worst This Century as Screen Time and AI Use Rise — Reuters |
