Generative AI in primary education: global perspectives

We are witnessing strikingly different approaches from countries around the world to incorporating AI into the classroom.

Some countries, such as China, are fully embracing the power of AI in primary education, while others, such as the UK, are engaged in trials to explore the impact that this technology might have. Norway has recently made the news for placing strict restrictions on AI use for primary-aged children.

Much of the consideration behind these different approaches comes from desire to develop children's cognitive abilities. As we will see, while countries are approaching this in different ways, all agree that cognitive development is a crucial part of primary education.

In this post, we will explore the distinct approaches taken by Norway, the UK, Estonia and China towards primary children using AI in education. The discussion will consider both the practical implications of each approach, as well as the research and philosophy behind the differing approaches of each country.

Norway - Guarding child development and the cognitive struggle

Norway's policy towards generative AI in education is rooted in cognitive psychology, as well as broader psychological considerations around use of technology by young children.

In 2025, the Norwegian Government published the white paper, 'Safe upbringing in a digital society,' (link)

This report laid out the government's objectives for keeping children safe in a digital society. It included considerations around screentime, privacy, online content, bullying and pitfalls relating to use of artificial intelligence by children.

On the topic of artificial intelligence (pages 19-20), the report did not raise the issue of use of AI by children lessening cognitive challenges. Instead, the report discussed how children can attribute emotions and intentions to AI tools. The report stated that parasocial relationships can develop as children bond with AI tools. It stated that children were vulnerable to manipulation by AI tools and that children had developed relationships with AI tools which were incapable of empathy or understanding.

On 19th June 2026, the Norwegian government published a press release, stating concerns that use of generative AI by primary children was affective their cognitive abilities. (link)

The press release stated,

"Several rounds of international surveys such as PISA and PIRLS show large drops in Norwegian students' basic skills, such as reading, writing and arithmetic. Uncritical use of generative AI can lead to students skipping important stages of learning."

The release went on to state,

"Our children must learn to find their own answers, not have them ready made by AI. They do not have the academic foundation, the critical reflection and self-regulation needed to use AI in a good way. They must learn to read for themselves, write for themselves, do math for themselves and think for themselves."

Coupled with this, the Norwegian government is also legislating to provide access to physical books. The press release stated,

"We are now sending a clear signal to municipalities that they must prioritise physical books to a greater extent."

The situation in Norway does not equate to a ban of AI in schools. As the press release stated, children in grades 1-7 will not be generally be given access to AI. In grades 8 - 10, AI can be used cautiously and gradually, depending on teacher competence. In upper secondary education, students will use AI appropriately to prepare them for further education and work.

In fact, the aforementioned white paper, 'Safe upbringing in a digital society,' stated,

"Artificial intelligence (AI) will play a significant role in pupils' future. Schools must prepare them for this...Schools must equip pupils for a future involving AI, enabling them to act with ethical awareness and critically reflect on the consequences of technology for individuals and society."

The UK - Piloting AI tools for personalised support

The UK Department for Education (DfE) is currently engaged in research and consultation on pupils using AI tools in the classroom. The DfE recognises the potential for use of AI in education, but places a high threshold to be overcome if AI tools are used, in terms of protecting students' privacy, data and cognitive work.

In the UK government policy paper, 'Generative AI in Education', updated 2025, (link), the government recognised,

"We have limited evidence on the impact if AI use on learners' development, the relationship of AI use and educational outcomes and the safety implications of children and young people using this technology in the classroom."

In the same paper, the DfE recognised that generative AI can be used for outcomes including,

• tailored feedback and revision activities

• supporting personalised learning

If pupils use AI tools, the DfE states that,

"Pupils should only use generative AI with appropriate safeguards in place, such as close supervision and the use of tools with safety and filtering and monitoring features."

Furthermore, in the the 'Generative AI product safety standards' (link), the DfE states that developers and suppliers should,

"Make every effort to mitigate the potential for cognitive deskilling, or long term developmental harm to learners."

In  'Generative AI in Education' (2025), the DfE outlines the piloting of an edtech evidence board, which is a group of experts who are assessing and evaluating the evidence and quality that edtech tools have a positive impact on teaching and learning.

Much of the guidance from the DfE seems to currently focus on efficacy of AI tools in schools by teachers, with more references currently to teacher outcomes in descriptions of research and pilot projects than use by students.

The National Literacy Trust carried out research (link) which details young people's experience of using generative AI for support with writing. This research drew some interesting conclusions, including that,

• Many young people used generative AI to support aspects of writing, from improving vocabulary (39.5%) to more creative or interactive purposes, such as asking for feedback on writing (20.7%) or to turn ideas into a story (19.3%).

• 1 in 5 (18%) young people who did not enjoy writing asked AI for feedback on their writing, indicating the potential of AI tools to support a broader range of learners

• 48.9% of young people who used AI added their own thoughts into it, 42.8% checked outputs as they could be wrong. 25.1% admitted to 'just copying' AI outputs when using it form homework.

The UK approach to children using AI in education is currently one of open minded research and data gathering, with an onus on providers of edtech products and ultimately schools, to check that AI tools do not harm students, including harming their cognitive development. As more data is gathered from focus groups and pilots, government policy and guidelines on provision of AI tools to students will develop further.

Estonia - Systematic adoption of AI tools for students

In contract to the UK's evolving strategy based on trials, input form expert panels and feedback from research organisations, Estonia's approach to AI is heavily formalised though a new, national strategy.

This strategy is called the AI Leap (TI Hupe) Initiative (link). The strategy, as reported by the European Commission (link), was designed to enhance personalised learning and improve efficiency in education. Under this strategy, in 2025, 20,000 students from grades 10-11 gained access to AI-powered learning applications. In 2026, this will be extended to vocational schools, giving an additional 38,000 students access to AI-powered learning applications.

To support this initiative, teacher training is already underway in Estonia and negotiations with Open AI and Anthropic have been initiated.

Interestingly, the cover page of the AI Leap website (link) recognises the risks to cognitive development that can be presented by use of AI tools. The website states,

"Delegating homework to machines is already common. Without intervention, the strongest would benefit from AI, but many students' critical thinking development would suffer."

The vision of the AI Leap program are stated as being,

• Together with teachers, students and parents, we shape an education where AI support learning rather than threatens it

• Estonian students and teachers have access to the best AI tools, as well as the skills and mindset for their smart and repsonsible use

• Estonia is a global leader in rethinking learning in the age of AI

Students participating in the AI Leap program will gain access to an AI learning platform developed specifically for Estonian students. This is described as being different from regular Chat GPT, in that it does not give direct answers, but acts 'more like a teacher', helping students plan their learning and formulate their own conclusions.

The AI Leap program currently provides children in grades 10 - 11 with access to AI tools for education. However, as the E-Estonia summary of the AI Leap program (link) outlines, the program will be evaluated in 2027-28 with possible extension for grades 9 and younger.

China - Cultivating early computational thinking

China's Ministry of Education (link) has stated an overarching goal of AI-powered education being to,

"Cultivate innovative talent capable of dealing with future challenges."

The Ministry of Education describes the need to "ignite students' interest in science by extensively involving them in project-based learning." The Ministry also describes the need to coordinate planning between different government agencies and to implement pilot projects.

The guidance provided by the Ministry of Education suggests that AI-related courses should be included at different educational levels as a curriculum component.

At lower primary-level, this should include raising awareness among students and at upper-primary level, the focus should be on understanding and using AI technology.

Significantly, schools are encouraged to incorporate AI teaching modules across the curriculum, including in science, IT, practical activity and extracurricular activities.

The guidance stresses that high-quality AI resources should be shared and developed and that AI laboratories at universities, research institutes and high-tech companies should be available to primary and secondary school students.

In addition to this, teachers should be recruited from universities, research institutes and high-tech companies to serve as part time teacher. Schools in rural areas should engage in teacher exchanges and should engage in interconnectivity between AI-related courses.

The Australian Government Department of Education (link) reports China's AI Education goals for primary level as being:

Cognition	Spark interest through hands-on experiences with smart devices and basic AI concepts like speech and image recognition
Skills	Develop basic AI skills by using simple tools, visual programming and practising data handling
Thinking	Build foundational thinking by practising logical reasoning, task breakdown, and comparing AI with human behaviour
Values	Deepen ethical understanding by recognising AI's strategic role in innovation and evaluating misinformation risks in generative technologies

China's goals for using AI in primary education include allowing children to understand the computational power of this technology itself, as well as benefitting from its outputs for their work across the curriculum. China aims to enthuse children about this technology and spark their curiosity and imagination, preparing them for innovative use of AI tools in secondary education and beyond.

Conclusion - Are we asking the right question?

As we look at the global landscape of AI in education, from Norway's careful consideration of cognition and development, to China's ambition for children's discovery and innovation, it is easy to frame the question as,

"Should children use AI in primary education?"

Instead, the differences in global perspectives show us that we must focus on the needs of the child and not on the technology itself.

The question we should ask is:

"What do primary children fundamentally need to develop into healthy, capable thinkers and what are the right tools to achieve that?"

When we reframe the question like this answers become clear, nuanced and actionable.

• Above all, children need to be safe and healthy. They should not be given access to tools that are not monitored and safe, whatever the perceived benefits.

• Children need foundational cognitive architecture. They need to learn how to solve problems logically and independently. They need to realise that struggle is a necessary component to development and they need to learn to appreciate and celebrate mistakes as opportunities to learn and grow. Children should have the chance to develop physically, develop fine and gross motor skills and develop strong communication and social skills.

• Children need equity and support. Children who have barriers to their learning should be provided with tools that enable them to overcome these barriers. These could be technological tools, educational resources, staff resources or taught strategies for coping.

• Knowledge is fundamental. Children need to be taught the value of knowledge as a foundation for creative self-expression. While technology can help children acquire knowledge, it should never be a substitute for children obtaining their own knowledge and appreciating the value of knowledge.

The differences in approaches to AI in primary education are notable in the four countries that we looked at. Significantly, all four countries mention cognitive development in the consideration of their strategy for AI in primary education, but approach this from different angles.

By framing the discussion around children's needs, rather than technological tools, countries can ensure that the approach they take is grounded in consideration of what children need to grow healthily and academically. It is likely that all countries could learn from the approach and rationale of other countries, and international dialogue and cooperation on incorporation of AI into primary education is therefore essential.