Cost of not investing in AI higher than investing in it, says Nandan Nilekani

Nilekani said he sees language AI as a crucial tool in bridging communication gaps, especially in a country with vast linguistic divergence.

Nandan Nilekani, the architect of Aadhaar and Infosys co-founder, has said the cost of not investing in artificial intelligence (AI) may prove to be costlier than investing in it, as companies could go out of business if there is a big platform shift over the next five years.

He was responding to a question on Zerodha co-founder Nikhil Kamath’s latest podcast if the massive AI investments could be a bubble, given the uncertainty over the revenue potential.

“They can’t afford not to do it (investment in AI), so they rather spend $50 billion a year than not spend… if there’s a big platform shift and they don’t have it, then they are out of the game permanently,” Nilekani said on Kamath’s podcast People by WTF.

Nilekani also touched upon the long-term impact of large language models (LLMs), saying it is already becoming a commodity. He said major tech players in the West are already spending billions due to a "competitive land grab situation".

LLM is a type of AI that can understand, generate, and respond to human language, which is trained on massive amounts of data and can perform tasks like answering questions and translating languages.

A commoditised product is one that is widely available and can be easily offered by competitors, resulting in sales depending on price rather than brand or features.

The collective annual investments worth $200 billion in AI infrastructure such as chips and data centres is a compulsion for Western companies. However, these heavy investments could ultimately lower the cost of AI and make it more accessible to India. "The good news for us is, all that is going to lower the cost and access to AI over time. It’s already happening," Nilekani said.

He said the launch of open-source AI such as Meta's LLaMA 3 is particularly beneficial to India since the country can leverage these advancements without the high cost of development.

Nilekani said he sees language AI as a crucial tool in bridging communication gaps, especially in a country with vast linguistic divergence. "If a farmer in a village in UP can speak in Hindi to the computer, and the computer can go and get some information on planting techniques to him and tell him in real-time back, suddenly we have made the world's knowledge accessible to that farmer in his or her language," he said.

His interest is in using technology at scale and for the benefit of the entire population, particularly for people who may not be literate or familiar with English. "If you can speak in your language of choice, your mother tongue, then suddenly you can access the entire knowledge in the system," he said, adding AI will help in expanding access to information.

The technocrat said AI is set to become as ubiquitous as electricity. "AI is going to be like electricity—it’s everywhere. So everything will have some AI component."

Nilekani also said his latest project is building something called Unified Energy Interface, where people can buy and sell electricity.

This infrastructure will be based on the open-source Beckn Protocol, backed by Nilekani, which provides specifications including APIs, data models, reference architecture, transaction mechanisms and global standards.

Beckn Protocol is the base for India’s Open Network for Digital Commerce (ONDC) and Namma Yatri, a ride-hailing service app.

“We are building something called Unified Energy Interface. People can buy and sell electricity. So anywhere that you have to buy and sell things between different people in a consistent way, Beckn is the way to do it,” he said.

The Infosys co-founder likened this idea of transacting energy to the way charging stations are used for electric vehicles. “I can go to a charging station and buy electricity through UEI or tomorrow somebody may say, I'll do a Beckn-based version of quick commerce, where instead of my investing in dark stores, I will use the retailers in my neighborhood, and I'll provide a quick commerce capability,” he said.

Also read: Working on a different architecture for electricity transactions, says Nandan Nilekani

When asked about which companies are doing well in the AI space and his investments, Nilekani said he was investing more for philanthropic causes than from investment point of view.

“I'm not really interested in the company so much. My interest is how do we use this stuff at population scale, low cost, to make a difference in education or health or farming or whatever,” he said.

Also read: Nandan Nilekani bats for data-driven small language models; bullish on Sarvam AI

In August, Nilekani said he is bullish on startup Sarvam AI, as it has put the world’s apprehensions to rest and built a small language model (SLM) with two billion parameters.

As part of commercial offerings, Lightspeed-backed Sarvam launched Sarvam Agents for enterprises, which are voice-enabled and multilingual custom business agents deployable via telephone, WhatsApp, or in-app.

Available in 10 Indian languages, the cost of these voice agents starts at Re 1 a minute, compared to a human agent costing Rs 10 a minute.