"LLMs Necessary But India Must Focus On Enterprise AI": Ajai Chowdhry To NDTV

India should strive to occupy the AI-for-enterprise space, which will help in areas like drug discovery and bring down research times and prices, Ajai Chowdhry, the Chairman National Quantum Mission of India's Mission Governing Board, said at the NDTV Ind.ai Summit.

Chowdhry also introduced quantum mechanics-based computing, which he introduced as "spooky physics", and said, "This whole technology is based on quantum mechanics."

This, he said, would add to current computing capabilities driving Artificial Intelligence and "provide phenomenal capability to do things we could not till now.... by classical computers".

"If we look at AI there has been a lot of talk about India being far behind in terms of LLMs, i.e., large language models. Now LLMs fundamentally use data that can be hugely taken from India or, and in addition, LLMs are fundamentally addressing consumer market, in a very big way."

Citing ChatGPT, he said, "This addresses consumer market in a huge way... I mean you ask ChatGPT anything and you get answers. But that is not the only area we should apply AI to."

"AI for enterprise, to my mind, is an area India should occupy. And if India occupies that area, where we use AI for helping enterprises to be successful in what they do today, that is where AI must be applied. Therefore, areas like pharma companies doing drug discovery."

"Today research for a drug typically takes 15 years and that is the reason behind cost of drugs today. We use a lot of generics here in India... but look at the cost of drugs in the US."

"Now, with AI drug discovery may get reduced by three to five years. You add quantum and you can reduce that by 6-7 years. Now look at impact on faster solving of problems of diseases ... because you can discover more drugs and reduce cost."

Investment in LLMs is necessary for "strategic autonomy", he said. But, at the same time, he warned against entering into discussions about India falling behind in the AI race.

Instead, he pointed to the SLM, or small language model, space.

"The 'are we behind' question... that is not a discussion to get into. We are not behind because we will get into SLMs in a big way. And, once we start implementing enterprise-based SLMs, then you will see the real result for corporates, MSEs, etc."