India should stop obsessing over generative AI, focus on sector-specific models, says IT secretary S Krishnan

IT secretary S Krishnan. (file photo)

• India urged to focus on sector-specific AI for productivity, beyond generative AI
• Smaller AI models can enhance healthcare, education, manufacturing, and agriculture.
• India aims to be a global hub for AI applications and open data collaboration

Did our AI summary help?

India must move beyond an excessive focus on generative AI and instead build smaller, sector-specific artificial intelligence models that can improve productivity in critical areas such as healthcare and education, information technology secretary S Krishnan said on January 30.

Generative AI is only one part of a much larger ecosystem, he said, warning against treating it as the sole benchmark of progress.

"Why are we so obsessed with the generative AI part? Why are we not looking at other aspects of it?" he said at the Democratising AI Access through Distributed Compute: Perspectives from the Global South event.

The event, an official pre-summit event of the India AI Impact Summit, was organised Deepstrat, Esya Centre, Qualcomm and AI Knowledge Consortium.

Previous generations of AI, when fine-tuned for specific tasks, can often deliver better and more practical outcomes. "These smaller models can make a significant difference in productive sectors of the economy — whether it is healthcare, education, manufacturing or agriculture," he said.

Krishnan also said India’s opportunity lay not in competing purely on large foundational models but in becoming a global hub for AI applications and use cases.

"We believe that India has the potential to be the application and use case capital of the world," he said, urging startups and technology firms to showcase solutions that can be deployed across the economy.

Krishnan outlined three pillars of India’s AI approach: infrastructure, models and data. On infrastructure, he stressed the need to create an open playing field for private players to participate in building and running digital and AI infrastructure.

This public-private model, he said, is increasingly seen by international organisations as suitable not only for India but also for much of the Global South.

On data, Krishnan highlighted the rapid expansion of government datasets on the AI Kosh platform. The number of open datasets has crossed 7,000 and continues to grow daily, he said, while acknowledging that large volumes of government data remain siloed and must still be opened up.

"Our job is far from being done," he noted, inviting private players to also contribute to the data ecosystem.