CHENNAI: Stressing the urgent need for India to build sustainable and sovereign artificial intelligence systems, V Kamakoti, Director of Indian Institute of Technology Madras, called for the development of indigenous AI models aligned with the nation’s values, strategic interests and long-term development goals.
Speaking at the 14th Think Edu conclave, Kamakoti said India must move beyond dependence on foreign AI systems and focus on creating Atma Nirbhar or self-reliant AI models. He said sovereign AI is not merely about reducing external dependency but about ensuring freedom from outside influence in shaping education, governance and public systems. “We must avoid a situation where external models influence how we think, learn or run our national processes,” Kamakoti said.
Speaking about Sarvam AI, he said it competes with some of the world’s most advanced frontier models and has established benchmarks for massive Indic language understanding.
Highlighting concerns around bias, security and strategic misuse, he outlined four critical questions driving India’s AI agenda: understanding model performance, ensuring long-term control, safeguarding intellectual property in design and infrastructure, and assessing potential harms in the Indian context.
Kamakoti underscored the importance of developing India-specific benchmarks, especially for Indic languages, noting that global benchmarks may not reflect the country’s linguistic and cultural diversity. He urged citizens to actively use and support Indian AI models so that these indigenous systems evolve with local data rather than enriching foreign platforms.
He also emphasised the need to build domestic computing capabilities from chip manufacturing to laptops and mobile devices capable of edge inferencing. “We cannot afford terawatts of power supply just to solve some stupid AI problem,” he said, linking AI growth to sustainable development .
His speech was followed by a panel discussion on the topic "How AI is changing education and what we need to know". In the interaction with author Sairam Surenderasan, Pawan Goyal, professor at IIT Kharagpur, emphasised that instead of using AI merely to make tasks easier, learners should ask whether it is helping them truly understand and enhance learning. The focus must shift from automation to skill-building developing the ability to acquire new skills quickly and effectively in an AI-driven world.
He also stressed the importance of sovereign AI models. Relying solely on global data can be limiting and potentially unsafe to achieve Atma Nirbhar (self-reliant) AI, India needs its own datasets. Collecting local data and feeding it into homegrown models, as being done in initiatives like Bharat Data Sagar, is crucial for building robust, contextually relevant AI systems.
In the context of using AI in education,Goyal said AI can enable personalised learning, tailoring content and support to individual student needs. However, it cannot replace teachers. The goal is to equip students with the mindset and tools to adapt to the evolving world, think critically, and leverage AI to enhance learning.