A global test of 760,000 teenagers has delivered a stark result: average reading and mathematics performance across OECD countries has fallen to the lowest level recorded in the history of the assessment.
The OECD released its 2025 Programme for International Student Assessment results on September 8, covering 15-year-olds in 91 countries and economies. It was the largest PISA assessment to date—and its clearest warning yet that the learning slide that began before the pandemic has not been reversed.
The numbers point to years of lost learning
Between 2015 and 2025, the OECD average reading score dropped 28 points, a decline the organization equates to roughly a year and a half of learning. Mathematics fell 22 points over the same period, or just over a year of learning. Non-OECD participants also recorded declines: 16 points in reading and eight in mathematics.
A separate Reuters analysis of the findings shows the longer arc. The average reading index fell from a high of 501 in 2012 to 466 in 2025. Mathematics dropped from 502 to 469 and science from a 2009 high of 506 to 486.
The problem is not limited to a small group falling behind. Across OECD countries, one in five students is now classified as a low performer in science, mathematics and reading simultaneously. That share was 16% in 2022.
Screens are part of the warning—not a single-cause explanation
The report connects weaker results with digital distraction, but the data show associations rather than proof that one device or app caused the global decline.
More than one in four students said classmates were distracted by digital devices in most or every science lesson. Those students tended to perform worse, show less engagement and feel less connected to school. OECD students reported an average of 3.4 hours of weekday leisure time on digital devices, in addition to roughly three hours spent using devices for schoolwork at school and home, according to Reuters.
Performance began to weaken when daily digital use became excessive. The study also found that nearly half of students regularly use artificial-intelligence chatbots for learning. Students who reported using AI for activities such as drafting essays, summarizing text and researching topics scored about 20 points lower in science on average—roughly the OECD’s benchmark for a year of learning.
That does not mean AI automatically reduces achievement. Students who lean heavily on chatbots may differ in other important ways, and digital tools can help when used deliberately. OECD Secretary-General Mathias Cormann said technology and AI can strengthen learning when they are used purposefully and do not replace attention, effort and understanding.
Reading for pleasure is disappearing
The assessment found that more screen time and less voluntary reading moved alongside weaker results. Researchers also detected more hurried reading: students were more likely to skim a passage, rush to an answer and miss its meaning.
That matters beyond literature classes. PISA is designed to measure how well students apply knowledge to unfamiliar, real-world problems. Reading comprehension supports performance in science and mathematics, especially when questions require students to evaluate evidence, identify misinformation or follow a multistep argument.
The decline was unusually steep among students from wealthier backgrounds. Socioeconomic advantage remains powerful—advantaged students still scored 85 points higher in science than disadvantaged peers—but the gap narrowed partly because higher-income students lost ground, not because lower-income students surged.
A small group of systems still leads
The highest-performing group was concentrated in East Asia. Participating Chinese jurisdictions, Japan, Korea, Singapore and Chinese Taipei were among the strongest systems across science, mathematics and reading. Estonia and the United Kingdom were the only countries outside that region to rank in the top 10 across all three subjects.
The OECD’s policy message is broader than banning phones. Its 377-page PISA 2025 report calls for deeper rather than more fragmented curricula, stronger investment in teachers, greater parent involvement and targeted help for the schools and students most in need.
What the results mean now
PISA is not a national curriculum exam, and country rankings can be affected by social conditions, participation and differences in school systems. But a decade-long decline across a sample this large is difficult to dismiss as statistical noise.
The most important finding may be the simplest: access to more information has not automatically produced deeper understanding. Schools now face a double task—teaching students to use powerful digital tools while rebuilding the concentration, sustained reading and independent reasoning those tools can too easily replace.
