What do educators and policymakers need to know from the latest PISA data?

The Programme for International Student Assessment (PISA) is a triennial measure of students’ ability to use their reading, mathematics and science knowledge and skills to meet real-life challenges. Conducted by the Organisation for Economic Co-operation and Development (OECD) it tests over 760,000 15-year-olds across 91 countries. The scale and methodological robustness of this study make the data some of the most important within education and highlight global and national trends. School leaders, as well as policymakers, have a responsibility to consider the findings and understand what implications they have for teaching and learning within their own settings.

England bucks the trend

Global standards remained relatively stable between 2006 and 2018 before seeing a significant decline between 2018 and 2022. The drop was steepest in mathematics and reading with only a modest fall in science. To some extent this was relatively unsurprising and the likely consequence of severely disrupted education during the pandemic, which manifested itself differently between countries but was universally detrimental. The 2025 data show a very steep decline in maths and reading scores, representing a 22-point drop and 28-point drop respectively in average scores across OECD countries since 2015.

By contrast, the outcomes for students in England significantly bucked the global trend. Scores in mathematics remained at 492, comparing very favourably with the OECD average of 465. In reading, the average score rose from 496 to 497, again looking particularly strong against an OECD average of 463. The impressive performance of students in England once again placed it ahead of the other UK nations. In both maths and reading, average scores declined in Scotland, Northern Ireland and Wales. While England, Scotland and Northern Ireland are above the OECD average, Wales falls below it, both in reading and maths.

In 2009, England was below Scotland in maths and only one point ahead of Northern Ireland. For reading, England was below both Scotland and Northern Ireland. Since then, England has consistently been the highest performing nation within the UK and the 2025 outcomes published this week have only extended the gap.

Education within the UK is devolved in Wales, Scotland and Northern Ireland, meaning that it is only English education for which Westminster and the Department for Education have responsibility.

The impact of artificial intelligence

The emergence of generative artificial intelligence between the previous PISA tests and the latest set, marks a significant change in the educational landscape. When analysing data, it is important to distinguish between correlative and causative relationships. Undoubtedly, the relationship between AI use and outcomes is complex but some emerging themes appear to be significant. Across the OECD, frequent use of AI for schoolwork was associated with lower science scores. The gap between students who did not use AI for the tasks examined and those who did was around 20 points. This is the equivalent to being ahead of one’s peers by around one year of schooling. However, the relationship varied significantly according to how AI was used. The difference in outcome is most stark when students use AI to draft text for writing assignments and even more so when they use it to summarise text that they were required to read.

The steep decline in reading across the globe, combined with the lower student outcomes associated with using AI to summarise text that should be read or to write responses, should be of real concern to school leaders and policymakers. Students who regularly use AI for general purposes and have learning opportunities at school to develop AI literacy skills, tend to have modestly higher scores in science. This means that how students use AI may be as important as whether they use it at all. The decline in deep and critical reading, alongside the general and sharp drop in reading attainment scores, should be a point of focus for educators. The growing use of AI may be one factor requiring particularly careful attention, alongside other technological distractions and changing reading habits. AI may also encourage “cognitive offloading” – the outsourcing of mental effort to technology – which may be particularly problematic for brains that are still developing.

Conclusions

It is important to pause and recognise the achievements of teachers, pupils, parents and policymakers in England. The nation’s strong and resilient performance – especially against a backdrop of substantial global decline – provides powerful evidence that many of the educational reforms and approaches pursued over the past ten to fifteen years have been beneficial. Those responsible for future policy should therefore be cautious about dismantling a system that has produced such strong outcomes.

In such a fast-moving technological landscape, which is transforming almost every aspect of human life, education needs to be research-led in its approach to technological use, and judicious and cautious in how students are permitted to use artificial intelligence. It will be important to help students understand that while it may provide shortcuts and efficiencies, there may well be long-term downsides, not just in terms of cognitive development but also in attainment data and examination outcomes.