Two-thirds of S’pore students use AI to help them learn at least once a week: PISA study

Sign up now: Get tips on how to help your child succeed

Nearly half of students here frequently use AI to research new topics, summarise texts and draft text for writing assignments.

Nearly half of students here frequently use AI to research new topics, summarise texts and draft text for writing assignments.

PHOTO: ST FILE

  • Two-thirds of Singapore students use AI chatbots weekly for learning, but those who don't use AI tend to score about 20 points higher in science.
  • Moderate and guided AI use can support learning, while overuse or lack of guidance may cause distraction and lower performance.
  • Family support for students in Singapore has declined, which concerns experts, as parental involvement strongly influences educational outcomes.

AI generated

SINGAPORE – Two-thirds of Singapore students reported using AI chatbots to help them learn at least once a week, compared with 46 per cent of students on average across Organisation for Economic Cooperation and Development (OECD) countries.

Nearly half – 41 per cent to 46 per cent – of students here said they frequently use AI to do research on new topics, summarise texts and draft text for writing assignments.

And only 4 per cent of students said they never or hardly ever use AI chatbots for schoolwork, according to a background questionnaire that was part of the Programme for International Student Assessment (PISA) 2025.

The questionnaire had posed 15-year-olds around the world a range of questions including their attitudes towards learning, how they used digital resources, how often they were bullied, as well as teacher support and parental involvement.

It surveyed 6,551 students across 157 schools here. The PISA 2025 results were released on Sept 8, and a record 91 countries and economies had taken part in the latest assessment, which examined the use of AI in school systems for the first time.

The study looked at the differences in AI use across high-performing systems. Singapore has explicitly included the use of AI tools, it noted, while the use of AI for schoolwork in Japan and Estonia is more limited in scope.

Students here reported spending on average 2.1 hours per day using digital devices at school for learning activities, compared with the OECD average of 1.7 hours. They reported spending one hour a day using digital devices at school for leisure, close to the OECD average of 1.1 hours.

In Singapore, 29 per cent of students said their peers are often distracted by the use of digital devices during most or all science lessons, similar to the OECD average (28 per cent).

More schools here have also introduced cellphone bans, with the proportion rising from 16 per cent in 2022 to 43 per cent in 2025.

On average across OECD countries, students in schools where cellphone use is not allowed on school premises are about 13 per cent less likely to report digital distraction.

The PISA report found a complex relationship between AI use and science performance, with the effects depending on how students use AI.

On average across countries, students who said they do not use AI outperform those who do in science by around 20 points – which OECD said is equivalent to being around one year of schooling ahead of their peers.

Speaking to The Straits Times on the findings, OECD’s director for education and skills Andreas Schleicher said AI separates task performance from learning, letting students “do wonderful things without learning anything”.

The report noted that poorer performance could be linked to other reasons, such as struggling students potentially being more likely to rely on artificial intelligence.

Students who do not use AI for specific schoolwork tasks outperform those who do. Meanwhile, students who use AI weekly for general learning perform similarly as those who do not use AI.

Students who report using AI chatbots to help them learn once or twice a week perform at a similar level fared by students who report not using AI at all, on average across the OECD.

This shows that digital tools can support learning when used purposefully and in moderation, said the report. In contrast, overuse or a lack of guidance can lead to distraction, limiting meaningful engagement and negatively affecting learning outcomes.

Students who frequently use AI for learning – and learn to assess the quality of AI-generated information at school – also achieve slightly higher science performance than non- or infrequent users, PISA found.

However, socio-economically advantaged students have greater access to these opportunities, raising concerns about a potential AI divide, the report noted.

Schleicher said the biggest surprise for him was how nearly 40 per cent of the young participants in Singapore said they would check information against sources but rely on common sense over scientific studies.

Students were asked about their views on whether information from a single source should be checked against other sources, and if they should rely more on common sense than scientific evidence.

“AI is common sense. AI is very fluent. It’s very persuasive. It’s very intuitive. In this age, I think common sense is no longer good enough. I think we need to actually reinforce that more scientific attitude, a shared reality that can be verified, not intuition and common sense,” said Schleicher.

He said the danger of AI is that “we give in our human autonomy, our ways of thinking, because technology provides easy solutions”.

Overall, as students engage less in advanced mental processes, this may also contribute to their decline in performance across all three areas in reading, mathematics and science, he added. “You do not get fit by watching sports – you get fit by doing sports. And the same is with learning; learning is always about cognitive struggle.”

Dip in family support

Despite strong teacher support in Singapore, students perceive their families as being less supportive in their learning, compared with their OECD peers. “That’s also a concern – that families increasingly become the client of education, not the co-producer,” said Schleicher.

In 2025, 55 per cent of students in Singapore said their parents or guardians asked them at least once per week what they did at school, while 37 per cent reported having at least weekly discussions about problems they might have at school.

“It’s not about making sure that your child gets into the best school,” he said. “It’s a lot more about what you do yourself as a parent to value the education. It’s not just about delegating that responsibility to an institution.”

“The PISA data shows that care and attention from families is a very powerful predictor for outcomes,” Schleicher said.

Acknowledging this finding, MOE said only about half of Singapore students reported having weekly family discussions about how they were doing in school, and 39 per cent said their families take interest in what they learnt in school.

“Children desire and benefit from such interaction with parents,” said the ministry, adding that its Parenting for Wellness resources can guide parents in supporting their children’s learning.

MOE also noted the study’s finding that Singapore students are also less physically active outside of school. While more students reported exercising outside school in 2025 than in 2022, 23 per cent reported not exercising at all outside school, higher than the OECD average of 17 per cent.

The ministry said it will continue to work with other agencies to help students lead an active lifestyle.

See more on