India needs its teachers to become social architects for AI adoption to work the way it should in classrooms

At its core, the challenge is threefold: we must scale access to high-quality learning, ensure depth of understanding and to develop judgement—and ultimately wisdom—in a world of abundant information.

The pedagogy risk: The global conversation on AI in education is converging on personalization—the idea of every child learning at their own pace through adaptive systems.

This vision is incomplete. Learning is not an individualized process but a social, cognitive act—shaped by explanation, disagreement and collective problem-solving. Personalization can support learning, but cannot replace the social processes through which judgement and collaboration are formed.

The incentives at work need attention.

Many digital systems maximize engagement rather than understanding, reward speed over reflection and prioritize ease over effort. Policy responses often focus on access rather than cognitive outcomes, creating environments that are stimulating but cognitively shallow.

So, as access to intelligence increases, the ability to think weakens. Since we risk cognitive dilution at scale, we must build systems where AI strengthens human cognition rather than takes its place.

An attention deficit crisis: Amid a cognitive debate, we face a structural challenge: the erosion of attention. Digital environments fragment focus and reward constant stimulation, weakening persistence and the ability to engage deeply with ideas. This is a neurological as much as a behavioural issue. If students cannot sustain attention, they cannot synthesize information or reach the level required for higher-order reasoning.

Education systems must therefore deliberately cultivate attention through sustained work, reduced distraction and structured reflection. We must treat attention as a core learning outcome that must be engineered into the school day. Without the ability to focus, the most sophisticated AI tools in the world will only serve to accelerate distraction rather than development.

From information to judgement and wisdom: We are entering a world where information is abundant and increasingly machine-generated. What is scarce is judgement, and ultimately, wisdom. The first develops through collaboration and adaptation to new contexts; in isolation, it could weaken.

Education has long been about content delivery, but must increasingly foster thinking skills. Instead of providing answers, it must shape how students question and decide. It must equip students with enough foundational knowledge to ask intelligent questions that drive deeper understanding.

Use AI for clarity, humans for learning: Early experiments by the eVidyaloka Trust point towards a different model. Their AI-enabled instructor, VidyaGanga, has been deployed across rural schools to teach in multiple Indian languages. Combined with their assessment engine, EduSprint, these systems deliver consistent explanations and immediate feedback. Early results indicate significant improvement in conceptual clarity.

Yet, their limits are equally clear.

These systems can explain and assess, but they cannot manage classroom dynamics, sustain collective engagement or push students to think through dialogue and disagreement. These capabilities remain fundamentally human. This requires a redefinition of the teacher’s role.

The teacher must transform from being primarily an information provider to becoming a social architect—one who designs learning environments for collective participation in thinking exercises.

The social architect could manage classroom dynamics with discussion and debate, shape group collaboration and introduce ‘productive friction’. This way, the teacher can sustain what no system can replicate: the social processes through which judgement, confidence and character are formed. In this model, AI ensures understanding, while the teacher ensures learning becomes thinking.

Re-anchor first principles: Meaningful integration of AI requires a return to first principles. Just as the Industrial Revolution led to the creation of gyms to preserve physical health, the AI era requires cognitive disciplines to preserve mental capability.

This includes writing (including by hand) as an exercise in thinking, as it remains essential for neural encoding; cognitive fluency in terms of an ability to recall say, math tables, as it enables higher-order reasoning; productive struggles, as effort results in learning and difficulty is a pathway to depth; social learning, as discussion and peer explanations drive deep understanding; and attention training, as uninterrupted work must be built into the school day.

An AI learning stack for India: The country needs digital public infrastructure to scale an AI learning model designed for our unique needs. This could include sovereign AI models aligned with Indian curricula and languages, open educational datasets as a public good and teacher-facing AI tools that support planning and facilitation.

Imagine a classroom three years hence where every student has access to high-quality explanations in their own language and assessments are continuous with instant feedback, but the centre of the classroom has shifted: students collaborate, debate interpretations and solve problems together.

The teacher orchestrates this process to ensure that learning moves beyond answers to understanding. Success will be measured by a student’s ability to sustain attention, explain ideas and question assumptions.

India has an opportunity is to build its own model. The future of education will be defined by how deliberately we design human learning, not by what technology has to offer. Staying relevant in an AI-driven economy requires us to re-imagine education.

These are the authors’ personal views.

The authors are, respectively, chief economic advisor, Government of India, and chairperson, eVidyaloka.