AI in South Asian Classrooms: Pendulums of Opportunities and Challenges by Tanvir Mostafa

AI in South Asian Classrooms: Pendulums of Opportunities and Challenges

Tanvir Mostafa

In education policy, artificial intelligence has gone from the periphery to the center. Ministries, school systems and teachers believe in AI’s potential for lesson planning, tutoring, assessment and managing schools. UNESCO now frames AI as a tool that might ultimately help solve some of education’s biggest challenges, accelerating progress toward SDG 4, but also cautions that the technology carries risks that are advancing faster than many regulatory systems. Its guidance on generative AI encourages countries to take a human-centered approach that protects privacy and designs age-appropriate use rather than addressing educational inequality as if A.I. were an automatic solution.

That tension is particularly evident in South Asia. The region has some of the world’s biggest school systems, some of its youngest populations and some of its deepest educational divides. AI is not arriving into a uniform landscape but rather to classrooms defined by multilingualism, uneven connectivity, teacher shortages, crowded schools and stark divides in urban and rural access. Thus, South Asia is in the throes of not just one, but several AI transitions simultaneously: India is transitioning toward deeper curricular integration; Pakistan and Sri Lanka are investing heavily in teacher and student AI literacy; Bangladesh and Nepal are testing readiness for learning interventions and teacher capacity; Bhutan and Maldives are building governance systems and digital literacies; while Afghanistan has an entrenched access–learning crisis.

What does AI, then, look like in South Asian classrooms today? In practice, the first wave is not so much about humanoid robots or fully automated teaching as it is about targeted tools. AI is being used, or widely discussed, for personalized tutoring, content generation, teacher support, automated feedback, translation and digital school management  and AI literacy. A recent program in Pakistan, supported by UNESCO, trained 150 teachers, 25 head teachers and 25 coordinators on using tools like Khanmigo for personalized support and lesson assistance. In the Maldives, the Ministry of Education has been developing a new education management system that it says utilizes AI to modernize governance. Bhutan National Education Technology Framework (BNEF) integrates connectivity, digital content, cybersecurity, and capacity building. The trend around the region is unmistakable. AI is showing up in education primarily via augmentation, rather than replacement.

India is perhaps the clearest case, thus far, of curricular mainstreaming. The CBSE also has dedicated set of AI curriculum resources for Classes IX to XII, where the Class X curriculum for 2025–2026 includes computer vision, natural language processing and Python along with project-based learning and explicit discussion of AI bias and ethical issues. India’s Ministry of Education similarly announced in late 2025 that AI and computational thinking would soon become part of the school experience for all children (starting from Class 3 on) embedding AI literacy much earlier into the educational journey. SriLanka is following a similarly complementary path, by having the country’s Ministry of Technology and Ministry of Education open AI clubs in schools as part of its National AI Strategy 2024–2028; the latter strategy emphasizes inclusive growth, public trust and ethical use. These are markers of nations treating AI as a civic and developmental issue, not just a technical one.

Bangladesh, Nepal and Pakistan showcase a second path: preparedness before scale. In 2024, the Asian Development Bank reported that Bangladesh was indeed laying the foundation for AI in Education via a needs assessment survey and a training workshop with primary teachers in (inclusive) 2023–24 on how teaching tasks could be supported by an increasingly powerful AI-enabled automation. The 2025 AI Readiness Assessment of Bangladesh (developed with support from UNDP) makes a direct connection between AI governance and Quality Education while outlining its pathways to ethical development. Education is among the sectors targeted to benefit from enhanced accessibility and efficiency through AI, according to Nepal’s 2025 National AI Policy, while a 2025 Nepal study revealed that there are many ways in which the integration of artificial intelligence in education is still at an infancy stage even as institutions, educators and policymakers actively explore implications on infrastructure and curriculum. Pakistan’s approach to training teachers shows the potential for classroom AI adoption to progress most quickly when education is prioritized.

Bhutan and Maldives model how smaller systems can leverage scale. Announced last November, Bhutan’s 2025 Education Technology Framework also aims to enhance learning through connectivity, digital content and teacher capacity building: launched in partnership with the EU and UNICEF by the Ministry of Education and Skills Development, it is expected to help over 6,000 students and 400 teachers in 10 pilot schools. Bhutan’s National AI Strategy similarly identifies education as one of its key focus areas and puts ethical, inclusive development at the heart of adoption. For example, the Maldives has linked education-sector transformation with AI governance at a national level. The Maldives’ education ministry announced an AI-enabled management system under the Thaiba project, and in 2025 UNESCO reported the completion of South Asia’s first AI Readiness Assessment by the Maldives. For a spread-island nation, AI-enabled officialdom and remote educational coordination may be particularly useful.

Afghanistan is different and any serious regional conversation needs to make that clear. In classrooms there, artificial intelligence remains more a future possibility than present reality. UNICEF’s education situation report for 2025 paints a stark picture of an unprecedented crisis: Millions of children specially girls are still excluded from education and learning outcomes are terribly  low. The report also highlights that over 90 percent of 10-year-olds are unable to read a simple text and states that teacher support programs employ mentorship, peer learning, and digital training modules. Given this context, the most urgent educational imperative is not automation in classrooms, advanced or otherwise; it’s restoring access, supporting teachers and protecting foundational learning. The lesson for Afghanistan is sobering: AI cannot mitigate systemic exclusion, and technological ambition must never supplant the right to education at all.

Those differences aside, the opportunities are deep. In South Asia’s multilingual classrooms, AI can also help generate explanations with varying complexity levels, translate or simplify content for understanding, or provide extra practice to students who are unable to keep up in crowded classrooms. It can reduce teacher workload by assisting in lesson preparation, question design, formative assessment and the provision of basic feedback. It can also improve overall inclusion through compatibility with assistive tech, especially in locations where specialist services are limited. In geographically fragmented systems such as that of the Maldives, or in remote and mountainous terrains in Nepal and Bhutan, AI-backed platforms can enable schools to connect with learners and teachers who are otherwise isolated. First and foremost, AI literacy is an essential core competency. Students should not just use AI tools; they need to scrutinize them, understand bias and think about when human judgment is more necessary than machine output.

But the threats are equally real. The first is inequality. AI tends to favor systems with devices, electricity, connectivity, reliable data and teacher time. The United Nations Educational, Scientific and Cultural Organization warns that failing to build inclusion into policy means AI will widen technological divides between countries and within them. South Asia’s own evidence corroborates that warning: Bhutan’s framework directly acknowledges infrastructure deficiencies and urban-rural divides; Nepal’s study highlights the issue of outdated syllabi, insufficient faculty training and infrastructural barriers; while Afghanistan’s crisis illustrates how poor infrastructure and exclusion can render even basic digital learning a challenge. A second threat is pedagogical: lean too hard on A.I. and critical thinking can atrophy, plagiarism becomes easier to commit and shallow learning rewarded if students earn credit for polished outputs rather than understanding. A third is ethical: child data privacy, opaque algorithms, language bias and unsafe or culturally inappropriate outputs continue to be unresolved issues across the region.

Sustainable adoption, then, needs a disciplined strategy. South Asian governments must start small with high-value use cases that combine teacher support, formative feedback, translation and accessibility, plus school administration. Before AI can improve learning outcomes, they need to invest in electricity, connectivity, devices and local-language content. Teacher preparation needs to be central, as opposed to peripheral; UNESCO’s AI competency framework for teachers is a helpful scaffold here because it treats AI literacy as a trajectory that leads from knowledge gain to the responsible practice of creation. Countries need child-sensitive data protection rules, transparent procurement standards and independent evaluation of tools, along with clear guidance on what students can and cannot use A.I. to do. Equally importantly, regional systems should privilege open, auditable and contextually relevant tools over proprietary black boxes that are alien to South Asian languages, curricula or classroom realities.

The broader message from South Asia is that AI in education isn’t just about a technology — it’s also about the governance of that technology, teaching and equity. The most promising models in the region don’t offer AI as a substitute for educators. Instead, they view it as a system of supports fashioned around curriculum assets, public values and local realities. If AI is merely used as a shortcut, it could entrench disparities in new digital ways. Yet, when embraced with care and attention to the teacher capacity, ethical guardrails, and inclusive design that is required of any meaningful transformation in education delivery models and systems, AI can be a powerful ally in creating classrooms that are more responsive, relevant and resilient across the region.

Tanvir Mostafa is a writer, poet, researcher and teacher based in Dhaka, Bangladesh, where he teaches English at Bangladesh Air Force Shaheen College Kurmitola. His works have appeared in journals such as Bhutanese Literature, Swapner Vela Sahitya Patrika, Dakshiner Darpan and so on admired for their depth and originality. His research articles often explore the intersections of literature, politics, and human nature, including studies on Shakespeare’s engagement with governance and power. He also researches Education, TESOL, ELT, and English Literature, examining how pedagogy and literary analysis intersect. Through both poetry and criticism, he explores nature, memory, and the resonance of human experience across time, making classical texts relevant to contemporary crises.