At a Global Technology Summit last week, Nandan Nilekani emphasized the importance of data over the obsession with models in the world of AI. The conversation mostly revolved around Digital Public Infrastructure (DPI) and Digital Public Intelligence to highlight the potential transformation shift for India’s technological landscape.
He said, “The data is what is missing right; we don’t have Indian language data, we don’t have audion data, we don’t have local data. So fix the data problem.”
Advocating for organizing data in a model-agnostic manner, harnessing open-source models, and fine-tuning them with indigenous data, predicting that within the next three years, models will become a commodity. He said, “Don’t waste your time on some model rat race.”
“Addressing concerns about restrictions on open-source models, optimism that the market forces would drive towards permissibility.”He highlighted “the significance of ensuring model’s permissibility for wider usage and competitiveness.”
Nilekani said, “I’m making a bet that open-source models will become more and more permissible. There’s also going to be competition. If I have two open-source models and one of them is permissible and one is restrictive, I’ll use the permissible one. Markets will take care of this. You wantr your model to flourish, then you please make it permissible.”
“The DPI journey was elucidated, umderscoring its profound impact on solving India’s multifaceted challenges. From the groundbreaking Aadhaar platform, providing identity to over 1.3 billion people, to enabling electronic KYC, which catalysed the surge in bank accounts through the Jandhan Yojana program.”
