Bhutan Expands Classroom AI as Students and Teachers Weigh Learning Gains and Risks

Grade 4 Maths Rollout Brings AI Into Focus

Bhutan is extending Eduten, an artificial intelligence platform for learning mathematics, to all Grade 4 students nationwide after a pilot supported by UNICEF showed an improvement of 11.2 percentage points in numeracy skills. The expansion puts a practical classroom programme at the centre of a wider debate: how can schools use AI to help children learn while protecting their ability to think for themselves?

The Ministry of Education and Skills Development (MoESD) is pursuing the rollout as students and teachers are already using tools such as ChatGPT, Gemini and Co-Pilot. These generative AI tools can produce responses to questions and requests, while specialised platforms such as Eduten focus on particular learning activities. Their growing presence is changing how students seek explanations and how teachers prepare, support and assess their classes.

As of June 9, 2025, the experiences described by Bhutanese students and educators show both enthusiasm and caution. Students value help with difficult subjects and heavy workloads. Teachers see opportunities to identify learning needs and reduce routine work, alongside concerns about copied assignments, inaccurate answers and the loss of independent thought.

What the Eduten Pilot Found

The Eduten pilot recorded high student engagement alongside the improvement in numeracy, the ability to understand and work with numbers. Its results helped support the decision to extend the platform to Grade 4 pupils across Bhutan.

Teachers involved in the pilot reported reduced workloads through automated analytics and personalised learning features. Analytics organise information about student performance, giving teachers a clearer view of where children need support. Personalised learning allows activities to respond to different learning needs rather than treating every pupil in the same way.

For teachers, those features meant more capacity to support students who were struggling. The reported benefit therefore reached beyond children completing exercises on a screen: information from their work helped teachers direct their attention. The nationwide rollout and alignment with the national curriculum form part of Bhutan's commitment to digital transformation while respecting local values and context.

Students Turn to AI for Explanations

Sherab Lhaden, a student at Motithang Higher Secondary School (MHSS), uses AI to clarify chemistry and mathematics concepts. She describes taking the information she receives and rewriting it in her own words, using the tool to support understanding rather than simply reproducing an answer.

Students across Bhutanese secondary schools also report using AI to verify answers and generate ideas. For some, its value is especially apparent outside school hours, when a teacher or classmate may be unavailable to explain a difficult point. AI gives them another place to turn when they become stuck.

A Class XI student at MHSS describes AI as helpful without considering it a daily necessity. That student also stresses the need to evaluate its responses critically. The experience reflects an important part of using these tools: students still need to judge whether an answer makes sense and whether they understand it.

At Lungtenzampa Middle Secondary School (LZMSS), students say AI can provide more detailed answers than a simple online search. Longer responses, however, do not always make a subject easier to grasp. Students there report that asking for further clarification can sometimes leave them confused.

Convenience Can Become Dependence

Sherab's appreciation of AI comes with concern about what happens when students rely on it too heavily. The MHSS student warns that easy access to answers can weaken the habit of working through a problem independently.

AI can make students overly reliant and less inclined to think independently,

Her concern is shared by peers who worry about the effects on memorisation, creativity and independent thinking. Those abilities matter both in examinations and in learning beyond school. A tool that helps a student finish an assignment quickly can also reduce the amount of thinking the student does if it supplies too much of the work.

Students also recognise that AI can produce inaccurate information. Their accounts show why detailed answers still require careful checking. Managing a heavy workload may make an immediate response attractive, but understanding the material requires students to examine the explanation and make sense of it themselves.

Teachers Look Beyond the Finished Assignment

Tshering Lham, vice principal of LZMSS, highlights the difficulty of monitoring AI use in schools and the urgent need for robust policies. Teachers are trying to explain responsible use while facing uncertainty about where assistance ends and replacement of a student's work begins.

Tshering Dorji, a geography teacher at MHSS, asks students to submit multiple drafts and gives feedback as their work develops. His approach encourages deeper engagement with the material rather than wholesale generation of an assignment by AI. Following the drafts gives the teacher a view of the learning process as well as the final submission.

Teachers are also encouraging students to use technology to refine original work. The central expectation is that the student's ideas and understanding remain part of the task. AI may help with that work, but the concern grows when the tool takes over its substance.

Large classes and heavy workloads make this difficult to monitor consistently. Teachers already responsible for many students must also consider whether polished assignments reflect genuine understanding or extensive assistance from AI.

Detection Tools Leave a Difficult Judgement

Teachers identify the lack of clear guidelines and dependable detection tools as a continuing concern. Students may be tempted to use AI to complete assignments without being detected, while schools need fair ways to assess the work submitted.

Recent studies described in the June 2025 account indicate that advanced tools such as Pangram can outperform human experts in identifying text generated by AI. Detection software nevertheless varies in effectiveness, and no solution is foolproof. A mistaken judgement can lead to a false accusation against a student.

Human oversight and AI literacy therefore remain essential to fair assessment. Teachers need to understand both how students use the technology and the limits of software intended to identify its output. The challenge combines questions about academic integrity with the practical difficulty of making sound decisions across large numbers of assignments.

Ethical Rules Need Classroom Attention

Students and teachers alike call for education about plagiarism, copyright and privacy. These issues reach beyond whether an answer is correct. They concern whose work is being presented, how material is used and what information is shared with a digital service.

Some schools have begun discussing such questions in IT classes, but comprehensive guidance is still lacking. No timetable for completing that guidance is given. Schools are consequently working through questions of acceptable use while AI is already part of students' academic routines.

UNESCO released AI competency frameworks for students and teachers in 2024. The frameworks aim to help countries such as Bhutan develop policies for effective and ethical integration of AI into education. Competency here means the knowledge and skills people need to use the technology responsibly, rather than simply knowing how to open an application.

For Bhutanese schools, ethical education and assessment rules are closely connected. Students need to understand what counts as their own work, while teachers need a shared basis for explaining expectations and responding when those expectations are not met.

AI Literacy Goes Beyond Asking Questions

Educators internationally are recognising the need to teach about AI as well as to teach with it. AI literacy includes understanding how to use a tool, questioning its responses and deciding when its assistance is useful.

Prompt engineering, the practice of crafting effective questions or instructions for an AI tool, is becoming an important skill for teachers and students. Recent research described in the account associates it with better instruction, reduced teacher workload and preparation for learning in a future shaped by technology.

The skill also connects with digital citizenship, metacognition and computational thinking. In plain terms, these concern responsible participation in digital spaces, awareness of one's own thinking and organised approaches to solving problems. Asking a useful question is part of a wider learning process, not just a way to obtain a longer answer.

Michigan's partnership with the AI Education Project offers an international example of efforts to expand AI literacy through teacher training and curriculum development. Bhutan's Eduten rollout provides its own practical setting for considering how technology, teaching skills and the national curriculum can work together.

Keeping Creativity and Local Values Central

Dorji Tenzin, a teacher at MHSS, warns that excessive reliance on AI can hamper problem solving, collaboration and creativity. These concerns extend beyond the accuracy of an individual assignment to the abilities students develop through doing the work.

Internationally, universities are experimenting with roleplaying bots powered by AI to help students practise skills used in real situations. The accompanying educational argument is that technology should support effective teaching rather than replace it. A useful classroom application still needs a clear learning purpose.

Bhutan brings its philosophy of Gross National Happiness (GNH) and its commitment to sustainability to this discussion. Thought leaders suggest that aligning AI policies with cultural values could help the country become a model for responsible, mindful technology use in education and beyond.

The practical task for schools is already visible in the experiences of their students and teachers: preserve access to helpful explanations and learning support while keeping original thought, creativity and human guidance central to education.

Key Points

• MoESD is rolling out Eduten to all Grade 4 students after a UNICEF supported pilot recorded an improvement of 11.2 percentage points in numeracy.
• Students use ChatGPT, Gemini, Co-Pilot and other AI tools to check answers, develop ideas and understand difficult subjects.
• Teachers and students raise concerns about inaccurate information, dependence on AI, plagiarism and the loss of independent thinking.
• Multiple drafts and teacher feedback offer ways to follow students' learning, while detection software still requires human oversight.
• UNESCO's 2024 competency frameworks provide a reference for ethical AI education, but comprehensive local guidance remains lacking.
• Bhutan's approach links digital learning with its national curriculum, cultural values, Gross National Happiness and sustainability.