AI in Education: Does It Improve Children’s Maths?

Key Takeaways

AI in education supported personalised arithmetic practice tailored to pupils’ needs.
Children using adaptive technology showed more positive arithmetic growth over three years.
Adaptive systems adjusted arithmetic exercises according to individual pupil performance patterns.
Adaptive technology helped teachers provide more individualised practice across classrooms.

For parents and teachers wondering whether AI in education can genuinely make children better at maths, the promise can feel both exciting and uncertain. What if technology could give every child maths practice that adapts to what they need, at the moment they need it? The idea sounds compelling, but does it actually translate into better learning when AI is used in real classrooms over several years? A large longitudinal study from the Netherlands brings that question into sharper focus, with findings that offer an encouraging but cautious answer.

The evidence comes from 7,885 children across 56 primary schools, where researchers compared pupils using AI-driven adaptive learning technology for arithmetic practice with those who continued their usual classroom curriculum. The results initially appear promising: children using the technology showed greater arithmetic growth on average. But the difference was not statistically significant, meaning the study cannot establish that the technology itself caused the additional growth. That tension between a positive trend and inconclusive evidence is what makes the research particularly worth examining, and it raises a deeper question about where AI in education can genuinely add value to children’s learning.

What kind of AI was studied?

The researchers studied AI driven adaptive learning technology, not generative AI such as ChatGPT/Claude. In general, Adaptive learning systems monitor how students perform and use that information to adjust subsequent learning activities. In this case, the technology could change the difficulty and sequencing of arithmetic exercises according to a child’s performance.

The goal is to provide practice that is better matched to an individual learner’s current ability. This matters because children in the same classroom can have very different levels of mathematical knowledge. A teacher can differentiate instruction, but doing so for every pupil can be difficult. AI driven adaptive technology offers one way to support that process.

What did the study find?

The children using the adaptive learning technology showed a trend towards greater arithmetic growth than those who did not use it. The researchers found small to medium effect sizes. However, the difference in long term arithmetic development was not statistically significant. That means the researchers could not confidently conclude that the observed difference was caused by the technology.

The study was also observational rather than a randomised experiment. Schools decided whether and when to introduce the technology. This means other differences between schools could have influenced the results. The researchers therefore describe their findings as an association rather than a causal effect. That distinction is important. It is not that AI in education has been proven to make children better at maths. It is that children using the technology showed a positive growth trend that deserves further investigation.

How does this compare with earlier research?

The findings are not necessarily at odds with previous research. A 2022 meta analysis examining 21 studies and 30 independent samples found a small positive effect of AI on mathematics achievement among elementary school students. The overall effect size was 0.351.

A broader 2024 meta analysis of 45 independent studies also found a statistically significant positive effect of AI enabled adaptive learning systems on learning outcomes. The overall effect was medium to large, with an effect size of 0.69.

However, the researchers behind the new Dutch study point out an important limitation of that earlier AI in education evidence: most of the studies included in the 2024 analysis lasted less than two weeks. The new research therefore addresses a different question on AI in education.

Do the potential benefits of adaptive AI continue over several years of normal classroom use?

The answer from this study is still uncertain. That does not make the earlier evidence irrelevant. Instead, the studies show why the duration and setting of AI use matter.

Could AI be particularly useful in some schools?

The researchers found some encouraging patterns when they looked beyond the overall result. Schools with a higher proportion of pupils from lower socioeconomic backgrounds showed greater arithmetic growth when using the adaptive technology. The technology also appeared to reduce some of the negative association between school size and arithmetic development.

These findings do not prove that AI works better in disadvantaged or larger schools. But they suggest that adaptive technology could be particularly useful where teachers face greater instructional demands and substantial differences between pupils.

That is consistent with the broader research evidence, which suggests that the effects of adaptive learning depend partly on the students, subject, duration and context in which the technology is used.

Did AI in education eliminate learning differences?

No clear evidence suggests that it did. The study examined individual characteristics including socioeconomic status and gender. Differences between students remained even when adaptive technology was used. That is an important reminder that personalisation is not the same as equality. AI in education can adjust the difficulty of an exercise. It cannot automatically address every factor that influences how a child learns.

Why does the teacher still matter?

This may be the most important lesson from the study. Adaptive technology can adjust practice according to student performance. It can also provide teachers with information about how pupils are progressing. But teachers remain responsible for interpreting that information and deciding how it should influence instruction.

The researchers emphasise that AI driven adaptive learning technologies are embedded in classroom practice. Teachers continue to provide instruction, feedback and guidance while orchestrating how the technology is used.

This means the potential value of AI in education may not come from replacing teachers. It may come from helping teachers do something that is otherwise difficult to scale: provide more individualised practice to many students at the same time.

What should parents and schools take from this?

Parents should not interpret the study as evidence that more AI based screen time will automatically improve a child’s mathematics. The research examined a specific adaptive learning technology used as part of classroom instruction. It does not tell us whether every AI tutor, chatbot or educational application will produce similar results.

For schools, however, the findings provide a reason to keep exploring the role of AI in education, particularly through adaptive learning. The technology may help teachers respond to differences in students’ abilities and provide more targeted practice. But its effectiveness is likely to depend on how it is implemented and how teachers use it.

Future research with more schools and greater control over technology adoption could provide stronger evidence about its long term effects.

What does this mean for the future of AI in education?

The evidence does not give us a simple answer about whether AI in education works or does not work. The broader research literature points to positive effects from AI enabled adaptive learning. Research specifically focused on elementary mathematics also suggests a small positive effect.

The new longitudinal study adds an important qualification. When AI driven adaptive learning is used in real primary schools over several years, the direction of the findings is positive, but the overall difference is not statistically conclusive.

That is not a reason to dismiss the technology. It is a reason to study it more carefully. The research suggests that the future of AI in education may be less about replacing teachers and more about giving them better tools for personalised instruction.

The promise of AI in education may not be that machines can teach children better than teachers, but that they can help teachers give more children the right practice at the right time.

FAQs on AI in education

Q: Does AI in education improve children’s maths skills?
A: Research suggests a potentially positive effect, but the evidence is not conclusive. The three year Dutch study found greater arithmetic growth among children using AI driven adaptive learning, but the overall difference was not statistically significant.

Q: What is AI driven adaptive learning in maths?
A: AI driven adaptive learning uses information about a student’s performance to adjust the difficulty and sequence of learning activities. In mathematics, this can help provide practice that is better matched to a child’s current level.

Q: How does adaptive learning technology help students learn maths?
A: Adaptive systems can adjust exercises based on how a student performs, allowing learners to receive more targeted practice. This may help teachers provide more personalised learning opportunities across a classroom.

Q: Is AI better than teachers for teaching mathematics?
A: The research does not show that AI in education is better than teachers. Instead, the evidence suggests that adaptive AI in education may be most useful as a tool that supports teachers with personalised practice while teachers continue to provide instruction, feedback, and guidance.

Q: What does research say about AI and maths achievement in children?
A: Earlier research has found positive effects of AI on mathematics achievement, including a 2022 meta analysis of elementary students. However, the newer three year study found a positive trend rather than a statistically conclusive overall effect, showing why long term classroom evidence matters.

Q: Can AI adaptive learning reduce differences between students in maths?
A: Not necessarily. The study found that differences between students remained even when adaptive technology was used, showing that personalising exercises does not automatically eliminate wider learning inequalities.

Q: Is AI adaptive learning suitable for primary school students?
A: The study specifically examined primary school children and found encouraging patterns in arithmetic growth among students using adaptive technology. However, results may depend on the technology, classroom setting, students, and duration of use.

Q: Does AI maths software cost more than traditional teaching?
A: The study does not establish whether AI adaptive learning is more or less expensive than traditional teaching. Cost effectiveness would require separate research comparing technology, implementation, teacher time, training, and learning outcomes.

Q: Should schools use AI for teaching maths?
A: The evidence provides a reason for schools to continue exploring AI adaptive learning, but not to adopt it simply because it is AI. Schools should consider the quality of the technology, how teachers will use it, implementation conditions, and evidence of learning outcomes.

Q: Can AI replace teachers in mathematics education?
A: Current evidence does not support that conclusion. The research points instead toward AI being used as a tool that can help teachers provide more individualised practice while teachers remain central to instruction and learning decisions.

References

Disclaimer:
Some aspects of the preparation of this content may be assisted by artificial intelligence or automated technologies and are subject to human editorial review and source verification. Readers are encouraged to consult the original research and primary sources for complete context. External links are provided for convenience, and TheHonores does not control or endorse their content. Relevant conflicts of interest, funding, sponsorship, or other disclosures are identified where applicable. This content is for informational purposes only and is not professional or medical advice. Images are for illustrative or representational purposes unless otherwise stated. Photo by Compare Fibre on Unsplash.