Cruising at the right speed on AIghway

Nirmala Sitharaman is promising AI entrepreneurs adequate policy, fiscal and credit support. Speaking at a technology institute in Kerala recently, she said that India is not just leading in the adoption of AI but also shaping how it is governed. Media reports suggest many companies have shown interest in building large and small language models. Ashwini Vaishnaw is confident the country will have its own AI foundation model this year. Domestic computing capacity is adequate for work on building large and small (sector-specific) language models. While having an indigenous foundation for AI is beneficial, the focus doesn't need to be concentrated on it. AI training hardware developed is now becoming a commodity. The value addition is in applications.

This is where India, with its deep pool of talent, can create a vibrant ecosystem that speeds up tech dispersal. Nandan Nilekani strikes the right balance in suggesting that India need not join the race for core AI models. His reasoning is to conserve computing power for use cases to bring down the cost of intelligence. Computing is energy-hungry, another constraint for a deficient nation. Cost-effective tech solutions have served India well in the past, and this could be the trajectory of domestic AI development as well. Tech dispersal in a price-sensitive market will depend on working around the computing and energy cost hurdles.

India stack of DPI creates a strong platform to deploy AI to solve population-scale problems. This is on display with Microsoft reporting that farmers in Maharashtra are making data-driven decisions for sustainable agriculture. India's large informal economy presents a bigger opportunity for AI deployment, which could offset reservations enterprises have to productivity enhancements. The pitch for India's role in AI development would be to train its sights on the use cases, while building core AI strength as it goes along. This should carve out a unique place for the country in the AI race.