PISA 2025 Reading Decline – A graph shared recently by The Economist carries a deliberately uncomfortable question: “Are teenagers growing dimmer?” The chart follows average PISA scores across a group of OECD countries over more than two decades, and the direction is difficult to ignore. Science has weakened moderately, mathematics has fallen more clearly, and reading has experienced the sharpest recent decline.
The question, however, should not be reduced to whether today’s teenagers are somehow “less intelligent” than earlier generations. The new OECD evidence tells a much more useful story about attention, reading habits, digital life, motivation and the ability to stay with a difficult problem long enough to understand it. That makes the findings important even for Sri Lanka, which was not among the 91 countries and economies participating in PISA 2025 and therefore does not have a directly comparable score in these latest rankings.
What Exactly Has Happened to Student Performance?
PISA, the Programme for International Student Assessment, tests 15-year-olds on how well they can apply what they know in reading, mathematics and science rather than simply repeat what they have memorised. More than 760,000 students across 91 countries and economies participated in the 2025 assessment, representing around 33 million 15-year-olds. The results released by the OECD on 8 September 2026 show that average performance across OECD countries has reached its lowest recorded level in all three core subjects.
Reading is the most concerning area. Across OECD countries, the average reading score fell by 28 points between 2015 and 2025, while mathematics declined by 22 points over the same period. OECD estimates that roughly 20 PISA points correspond, very approximately, to one year of learning for a typical 15-year-old, meaning that the scale of the reading decline is educationally substantial rather than a small statistical movement.
| PISA measure | 2015–2025 change across OECD | What it suggests |
|---|---|---|
| Reading | -28 points | Roughly 1.5 years of learning by OECD’s broad benchmark |
| Mathematics | -22 points | A little over one year of learning |
| Science | -7 points | Smaller decline over the decade |
| Low performers in all three subjects | 20% in 2025 | Up from 16% in 2022 |
These figures should not be interpreted as proof that young people have become less intelligent. PISA measures a particular set of abilities under test conditions, including reading carefully, interpreting information, solving problems and applying knowledge. What is changing may therefore tell us as much about how teenagers learn, focus and interact with information as it does about what schools formally teach.
PISA 2025 Reading Decline: The Most Disturbing Finding Is Not the Score
The deeper OECD analysis makes the reading results more interesting and more worrying. Between 2018 and 2025, the proportion of students classified as accurate and fluent readers declined, while the share of “hasty readers” increased. These are students who respond quickly but incorrectly, suggesting that part of the decline involves difficulty maintaining the careful attention required for demanding texts.
The pattern becomes stronger later in the reading assessment, when concentration has to be sustained for longer. OECD researchers found that students were increasingly likely to provide quick and superficial answers rather than spend more time working through difficult material. Their conclusion is careful: the evidence does not show that teenagers suddenly lost the ability to read, but it does suggest a weakening capacity or willingness to sustain attention during more demanding reading.
This matters far beyond English lessons or literature classes. Reading deeply is how a student understands a complicated science question, follows a mathematical problem, compares conflicting claims, interprets a contract or identifies whether something circulating online is misleading. In the age of generative AI, deep reading may actually be becoming more important precisely because machines can now produce enormous amounts of convincing-looking information almost instantly.
Screens Are Part of the Story, but Not the Whole Story
It would be tempting to look at the results and simply blame smartphones. The evidence does not support such a simple conclusion, and the OECD itself warns against assuming a direct cause from correlation. The decline in reading had already begun before the COVID-19 pandemic, while changes in reading habits have been developing for many years as young people spend more time encountering shorter and more fragmented forms of digital text.
The pattern is still difficult to dismiss. OECD data show that students increasingly read through online searches, messaging and social media while reading fewer printed books, magazines and newspapers. Earlier PISA research also found that students who regularly read longer texts and fiction tended to perform better in reading even after differences in socio-economic background were considered.
Digital distraction inside schools creates a separate concern. In PISA 2025, 28% of students across OECD countries said classmates were distracted by digital devices during most or every science lesson, while students using digital devices heavily for leisure at school tended to perform worse. Schools with clear mobile-phone restrictions or formal rules also tended to report lower levels of distraction.
The lesson is not that education should return to a world without technology. Digital tools and AI can support learning, research and personalised practice when they are used deliberately. The real issue is whether technology is helping students think or constantly giving them another opportunity not to.
The Countries at the Top Give Us Another Clue
A recent ranking of countries by reading performance using PISA 2025 data places Singapore at the top with a reading score of 535, followed by the four participating Chinese jurisdictions at 527. Chinese Taipei scored 508, Japan 503, while Macao and Korea were both around 501. Ireland was the strongest-performing European country in this comparison at 500.
| Selected high-performing systems | PISA 2025 reading score |
|---|---|
| Singapore | 535 |
| B-S-J-Z (China) | 527 |
| Chinese Taipei | 508 |
| Japan | 503 |
| Macao (China) | 501 |
| Korea | 501 |
| Ireland | 500 |
| Estonia | 499 |
| OECD average | 461 |
The table should not be treated as a simple recipe where Sri Lanka copies Singapore or Japan. Education systems operate in different languages, cultures, income levels and social conditions. What matters is that several high-performing systems combine strong academic foundations with structured teaching, teacher investment and comparatively controlled classroom distraction, showing that technological development does not automatically require weaker traditional skills.
Sri Lanka Has an Important Problem: We Cannot Make This Comparison
The absence of Sri Lanka from PISA 2025 is itself worth discussing. We cannot look at the latest international data and say whether a Sri Lankan 15-year-old is reading better or worse than a student in Singapore, Estonia, Türkiye or the Philippines because Sri Lanka does not have a score in this assessment. That means we should resist both pessimistic claims that our children are definitely experiencing the same decline and comfortable assumptions that the problem does not apply here.
Sri Lanka has examinations, school assessments and national education data, but international assessments such as PISA answer a slightly different question. They test whether students can take knowledge learned in school and use it to understand unfamiliar problems, evaluate information and reason independently. In a world of AI-generated answers and endless online information, having a clear national measure of those skills would be increasingly useful for education policy.
Participation in an international assessment should therefore be considered not because Sri Lanka needs another ranking to celebrate or fear, but because policymakers need comparable evidence. Without measurement, debates about declining reading, attention, digital learning or critical thinking can easily become arguments built from personal experience rather than national data.
This Should Not Become a War Between Books and Technology
Sri Lanka should also avoid the wrong reaction. Replacing every screen with a textbook would be as simplistic as replacing every textbook with a tablet. The OECD findings suggest that purpose, amount and context matter more than whether a device is present at all.
Schools could protect daily time for uninterrupted reading, especially longer texts that require students to follow an argument over several pages. Teachers can deliberately ask students to explain what they have read rather than simply locate an answer, while libraries need to become active parts of school culture rather than rooms students occasionally visit. Parents also matter because OECD research has linked children’s enjoyment of reading with seeing adults read and value reading at home.
AI requires similar discipline. A student who asks an AI tool to summarise every chapter, solve every problem and prepare every written response may complete more work while practising less thinking. Used differently, the same tool can explain a difficult idea, generate practice questions or help a student compare their reasoning with another explanation, making the difference between substitution and support extremely important.
The Real Question Is Whether Our Children Can Still Stay With a Difficult Idea
The image asking whether teenagers are “growing dimmer” attracts attention because the headline is provocative. The evidence underneath it deserves a calmer but more serious response. PISA 2025 shows a long decline in reading and mathematics across many developed education systems, and the fact that reading continued to fall even after the pandemic period makes it increasingly difficult to explain the problem through school closures alone.
For Sri Lanka, the warning should arrive before we have our own crisis graph. Education policy has understandably become interested in digital classrooms, technology, AI and preparing children for future jobs, but none of those ambitions removes the need for a teenager to sit with a difficult page, understand it, question it and connect it with something else they know. A country that loses that ability may produce students who can access more information than any generation before them while becoming less able to judge what that information means.
That is why the most important response is not nostalgia for an age before smartphones. It is rebuilding a culture of attention around modern technology, measuring whether students can actually understand what they read and ensuring that digital tools strengthen rather than replace the difficult mental work through which learning happens. If international scores are telling us anything, it is that being surrounded by information is not the same as learning how to think with it.
For further reading, see Ceylon Public Affairs’ analysis of Sweden’s education rethink and reduced classroom screen use and our broader examination of Sri Lanka’s education system. Both debates connect directly with the question raised by PISA: how education systems can use modern tools without weakening the foundational skills on which every later form of learning depends.
This analysis is based on OECD PISA 2025 results released on 8 September 2026, the OECD performance-trend data reflected in the accompanying Economist visual, and the Visual Capitalist reading comparison published in September 2026. Sri Lanka did not participate in PISA 2025, so no claim is made about Sri Lankan student performance relative to the countries ranked.
