While Srinivas commended Nilekani’s extensive contributions through Infosys and UPI, he disagreed with his stance on training AI models, urging India to focus on developing its own AI models
The rapidly advancing field of AI and Machine Learning is sparking intense discussions within India, with industry leaders offering differing perspectives on how the country should navigate its AI journey. One of the most significant points of contention involves the decision to either focus on building indigenous Machine Learning Models or to rely on improving and customizing existing models. Two key figures, Aravind Srinivas, CEO of Perplexity AI, and Nandan Nilekani, Co-founder of Infosys, are at the forefront of this debate, each advocating for a different approach.
Also Read: PM Modi discusses evolution of AI with Aravind Srinivas, CEO, Perplexity AI
The Perplexity CEO has strongly criticized Nilekani's suggestion that Indian AI startups should avoid creating large language models (LLMs) and instead focus on practical AI applications. While Srinivas commended Nilekani’s extensive contributions through Infosys and UPI, he disagreed with his stance on training AI models, urging India to focus on developing its own AI models. He emphasized that both model optimization and building upon existing models are crucial to the country’s long-term AI capabilities. According to Srinivas, India's AI future should mirror the achievements of ISRO in space exploration, where the country demonstrated its ability to develop world-class technology with relatively modest investments.
Call to strengthen India’s AI foundations
Srinivas shared his experiences running Perplexity AI, highlighting the mistaken belief that building foundational AI models would require vast amounts of money. He argued that India’s AI community could succeed in this area with a more focused effort, citing Elon Musk's admiration for ISRO’s achievements as a proof of concept. Musk, who recognized ISRO’s ability to deliver impressive results on a budget, serves as an example for how India can approach AI and Machine Learning development efficiently.
Additionally, Srinivas called for India to take a more proactive role in creating AI models that are competitive on the global stage. While acknowledging that India's focus on Indic languages is important, he stressed the need for India to create AI models capable of competing on all global benchmarks.
He referenced the recent success of Chinese AI company DeepSeek as an example of what India could achieve if it focused on training AI models. While Srinivas expressed that he wasn’t in a position to spearhead such a business, he offered his support to anyone willing to take on the challenge, even suggesting that the resulting models be open-sourced to benefit the wider community.
Also Read: Beckn And Finternet Together For Global Climate Impact: Nandan Nilekani
