AI Isn’t Enough to Fix Indian Schools

Schools worldwide are racing to integrate AI into classrooms. India is no exception, with states and education boards increasingly embedding AI into school curricula.

Punjab recently became the first state to introduce AI as a mainstream subject from Classes I to XII under its Sarbat.AI initiative, with the rollout taking place in phases across government schools. 

Kerala, meanwhile, was among the earliest adopters, embedding AI concepts into its ICT curriculum through Kerala Infrastructure and Technology for Education (KITE), exposing students to machine learning, AI ethics and practical applications.

At the national level, the Central Board of Secondary Education (CBSE) has also expanded AI education. Beginning with the 2026-27 academic session, CBSE introduced a new Computational Thinking and Artificial Intelligence (CT&AI) curriculum for Classes III to VIII, helping students build logical reasoning, problem-solving skills and AI literacy from an early age.

Yet, as AI makes its way into classrooms, some educators argue that India’s biggest education challenge lies elsewhere. Before AI becomes part of everyday learning, they say, every child must first be able to read, write and build strong foundational skills.

“Our problem in India is very basic. We need to push children who don’t have access to quality education first to a certain level so they don’t drop out of school,” says Shekar Sivasubramanian, Head, Wadhwani AI, in an exclusive interview with AIM.

Sivasubramanian says scaling AI-led education initiatives across India is challenging due to the country’s linguistic diversity and decentralised education system, which makes it difficult to adopt a one-size-fits-all approach.

“India has additional complexities because of its multilingual and significantly diverse nature as a country. It’s very difficult to have a single rule that will be applicable in Meghalaya, Tamil Nadu, Punjab and Gujarat,” he says.

For Sivasubramanian, AI literacy is less about coding and more about application. He argues that AI should be integrated into the formal curriculum in a way that teaches students how to apply it within their chosen disciplines. 

“Nobody needs to immediately become an AI programmer. They don’t need to learn logistic regression or neural networks. They need to become users and understand how AI helps solve problems,” he says. 

Wadhwani AI, in collaboration with The Bridgespan Group and supported by Google.org, recently released a report outlining a roadmap for integrating AI into India’s public education system.

India has more than 250 million students across 1.5 million schools, with nearly 60% enrolled in government and government-aided institutions. According to the report, AI can help improve learning outcomes, increase system efficiency and strengthen links between education and employment through personalised learning, multilingual access and career guidance.

The report estimates AI-enabled education could add $2.6 trillion to India’s GDP by 2047 and lift annual GDP growth by up to one percentage point, supporting the Viksit Bharat 2047 vision.

Basic Foundational Skills are Necessary

Wadhwani AI believes India’s biggest education challenge is far more fundamental, with millions of children still struggling to read fluently.

The organisation is tackling this challenge with its solution called the Oral Reading Fluency solution, which has already reached 8.7 million students across Gujarat and Rajasthan. 

The platform is intentionally simple. Students are shown an age-appropriate paragraph in their native language and asked to read it aloud into a mobile phone. Behind the scenes, AI evaluates pronunciation, fluency, skipped words, incorrectly spoken words, and overall reading ability.

Instead of manually assessing an entire classroom, teachers receive grouped insights that classify students into different learning cohorts. Children who cannot read at all or struggle to form sentences are placed in the lowest cohort, while fluent readers fall into the highest category.

“What the teacher wants to know is which groups of students need attention,” Sivasubramanian says. “If we can move children from the weakest cohorts into stronger ones, we’ve significantly improved foundational literacy.”

The application currently supports English, Hindi and Gujarati, with additional languages including Odia and Telugu under development. “Our goal by 2027 is to reach at least five crore children,” Sivasubramanian says.

Besides that, Wadhwani AI also offers Spoken English Assessment and Practice (SEAP) to help college students and learners from regional language backgrounds improve their spoken English through AI-driven assessment and practice.

Why AI Can’t Replace Teachers 

Unlike many AI education products that attempt to answer student questions or explain concepts, Wadhwani AI deliberately avoids acting as a tutor.

“This tool is not meant as a learning assistant,” Sivasubramanian says. “It is an assessment and remediation tool.”

The organisation believes teacher-led classrooms remain the most practical model for India’s public education system, particularly in rural schools where device ownership and internet access remain limited.

“In many rural households, there may only be one phone, which is usually with the male member of the family. The child is often the lowest priority. These are social changes that will take time.”

As more families gain access to smartphones, Wadhwani AI expects parents and teachers to encourage children to practise at home. Until then, the classroom will remain the primary place where the technology is used.

AI Will Change Jobs

Beyond school education, Sivasubramanian says India’s education system will also have to prepare students for a workforce increasingly shaped by AI

“Our education system will have to move towards much higher skill-based education and cognitive capability,” he says.

He expects routine knowledge-processing roles to shrink dramatically over the next few years. “The world of people who primarily take information, assess it, summarise it and transform it will disappear.”

Moreover, Sivasubramanian says the focus of education needs to shift from simply securing a job to building genuine expertise in a subject.

He illustrates the point with software development, saying experienced programmers with strong fundamentals can generate enormous amounts of high-quality code using AI, while weaker programmers struggle despite having access to the same tools.

The solution, he argues, is not to compete against AI but to master one’s discipline. “When you know your subject very well, AI becomes your slave. If you don't know it well, you become an AI’s slave.”