Nandan Nilekani Praises DeepSeek, Critiques Complex AI Models

Infosys co-founder Nandan Nilekani recently criticized the growing trend of developing increasingly complex AI models, arguing that these efforts are producing diminishing returns, according to the sources.

Speaking at the People+AI Mela in Bengaluru, Nilekani stated, "They keep pushing for more, but it’s not necessarily improving. These models are often confined within the walls of a company, acting as black boxes to the rest of us." He stressed that the lack of transparency and accessibility stifles broader innovation and progress.

In contrast, Nilekani praised China’s innovative approach to AI model development, noting their ability to create efficient models with fewer resources. He commended China’s transparency, saying, “They have done the world a service. This goes back to commoditization.” Nilekani believes that China's willingness to share techniques has played a major role in the commoditization of AI knowledge, benefiting global AI progress.

Nilekani has long viewed AI models as commodities, stating, “This is proving to be true. Hundreds of models are being built. China’s DeepSeek has demystified model development. Just last week, they shared all their open-source practices. The knowledge to build models is rapidly becoming commoditized.” He sees this commoditization as essential for driving innovation and making AI technologies more accessible.

Emphasizing the importance of computing power and data, Nilekani underscored the need for high-quality datasets, especially in Indian languages, to accelerate AI development.
He pointed to AI4Bharat, an initiative by IIT Madras focused on open-source language AI for Indic languages, as a prime example of how local data can be harnessed to drive AI progress. “The hard part is uses. How do we use this to deliver value for people? That is where we can lead. We can always replace the model.”

Supported by Nilekani for the past three to four years, AI4Bharat is focused on expanding its efforts over the next three years. The initiative will showcase how AI becomes an accessible and important resource in every vernacular from diverse landscapes in India.

India's entrepreneurship ecosystem has witnessed a phenomenal growth of startups- from a mere figure of about a thousand in 2016 to about 150,000 in the recent years. Nilekani also envisages further growth as the result of wide-spread mass-scale technologies combined with an increase in capital investment. “They will use mass scale technologies that we have deployed in interesting use cases. There is also going to be a lot of capital coming in,” he noted.

Nilekani has stressed that real innovation in AI applications is where the real impact can be created for ordinary people. While AI models might keep improving, attention should be paid to how well AI would effectively solve real-world issues and enhance living standards.