Our Terms & Conditions | Our Privacy Policy
The time is ripe for India’s “UPI moment” in artificial intelligence, and a comprehensive AI stack built as a digital public infrastructure (DPI) will help take AI to the masses, industry leaders and policymakers said. Speaking at the third edition of The Economic Times Digital Transformation Dialogues in New Delhi, experts sought regulations that would ensure trust and accountability without stifling innovation. While concerns on job displacement remained, education and entrepreneurship could be a solution to combat this, they said in a conversation moderated by ET’s Dia Rekhi.
“We need to, at some point, think of developing a full AI stack as a DPI. I think then we can see a real proliferation of AI to the end users in India,” said Rajendra Kumar, secretary (border management) at the Ministry of Home Affairs (MHA). He noted that such a model would empower both startups and government departments to develop use cases and launch applications faster, without the burden of building their own compute or training their own models.
Sunil Gupta, cofounder and CEO of Yotta Infrastructure, agreed: “We are all waiting for the UPI (unified payments interface) moment to happen in AI now. All the enabling factors in India are there.”
He pointed out that India has low-cost and abundant graphics processing units (GPU) infrastructure for startups and researchers for model training and to trigger an “inferencing wave”. Inference is the process of an AI model making a prediction or solving a task based on data it was trained on.
“AI use cases, if you are able to bring to Indian masses at no or very low value – that is how we consume as Indians – it is going to proliferate at population scale,” Gupta said. Such a time is possibly “coming very fast” as AI initiatives in areas like healthcare, agriculture and education are already underway, he added.
Live Events
Era of agentic AI
Discover the stories of your interest
Among the most talked-about trends, according to the panellists, is the emergence of agentic AI – systems capable of making autonomous decisions.”The hype (about agentic AI) is absolutely real and it is getting better and better every day,” said Sanket Atal, managing director of Salesforce India operations. “Today, we do more than a trillion transactions – whether it’s predictable or generative-every week around the world. With that came copilots, etc. And today, we are in the era of agentic. Agentic just ties everything together.”
Atal emphasised that the core of AI remains data, which is why Salesforce is investing in tools like its Data Cloud, which enables users to bring all their data together. “That establishes an awesome foundation to create agents and have everybody leveraging AI,” he said.
Besides enterprises like Salesforce, many startups, too, are betting on agents. Ankush Sabharwal, founder and CEO of Corover AI, said developers using its BharatGPT platform have already created around 3,500 AI agents. However, he cautioned against adopting AI agents blindly: “We should not have FOMO (fear of missing out) of using and force-feeding the new technology. Currently, the AI agent cost is more than the actual physical agent.”
While more than 1,000 users have shown initial interest, Corover AI has only about 100 managed accounts. “They’re not paying because there is no immediate value,” Sabharwal said.
The panel also debated whether India should build AI capabilities from scratch or on top of existing applications.
Sachin Bhatia, cofounder of cloud communication platform Exotel, said India should focus on building companies that enable AI. “I would love an Nvidia to happen out of India – now, that would be of value,” he said. “Ultimately, the winners are the people who own the infrastructure, or who own the application – everything in the middle is almost free now.”
To regulate on not
Regulation of AI was a key theme, with consensus leaning toward ‘light-touch’ frameworks.
Sabharwal argued that lack of regulation has hampered adoption. “People are not able to put millions of dollars in somewhere where policies are not there,” he said.
Ashish Aggarwal, vice president and head of public policy at Nasscom, underscored the need for clear regulatory intent. He advocated looking at achieving desired objectives as well as ethical AI practices considering the entire value chain across AI developers, deployers, clients and end customers.
“Responsible AI is driven by the very real business need to get things right, and also to avoid any liability,” Aggarwal said. “One of the things that we are now working with the industry on is to say that once you put out something, then how can you effectively demonstrate that you are living those values and living those governance principles?”
Bhatia of Exotel offered a simple principle for companies to stand by – never try to fake that an AI agent is a human being. “If it’s AI, it should be known that it’s an AI,” he said. “The second thing is, building observability within any use case, which means that what the AI did should be transparent – you should have a log of what happened so that you can trace it back,” Bhatia said,
Atal highlighted how Salesforce builds the concept of trust in their AI products. There are guardrails to contain hallucinations and toxicity in large language models (LLMs). “We enable the users of our technology to actually keep their own data private and not have that be trained in the LLM,” he said.
Impact on jobs
As AI gets more sophisticated, it’s bound to impact more and more jobs, panellists agreed.
“There are not going to be that many jobs. That is very clear,” Exotel’s Bhatia said. “I want us to focus on that problem from a regulatory or a government (perspective). For example, if you are making people redundant, how do you make sure that their wellbeing is taken care of over some time?”
While new roles will emerge, he warned they won’t fully offset the old ones.
Nasscom’s Aggarwal pointed to education reform and fostering an entrepreneurial mindset as long-term solutions.
“What is happening is some tasks are getting automated… And given the global need for stuff and India being the supplier of services, we don’t have a demand side constraint. In that sense, the short and medium-term story is actually good for India,” he said. “In the long term, we don’t know what is there, but it’s a topic worth really thinking about.”
Despite the challenges, the panellists remain optimistic.
“Huge gains and amazing economic impact can be had by adopting AI,” Salesforce’s Atal said. “I don’t think everybody needs to become an AI expert. The whole concept of being able to design systems is to enable people to use the technology for their own betterment.”
Dilip Kant Jha, chief information officer of SLMG Beverages, the largest Coca-Cola bottling company in India and South Asia, concurred. He said his firm has started using agentic AI to streamline logistics, predict safety incidents, optimise inventory and prevent overproduction.
“It gives you real-time visibility – looking at the market trends, it will tell you what to produce, what to send, what stocks you should keep – it is a gamechanger,” Jha said.
However, there should always be a human in the loop and a focus on ensuring datasets are not biased, he added.
Privacy safeguards are vital in AI deployment but exceptions may need to be made in cases where the technology is used to strengthen national security interests, MHA secretary Kumar said.
He stressed the need for carve-outs for exceptional cases “where we need to really ensure that the threats are detected real-time and then we can respond to them as quickly as possible.”
AI tools can enable more efficient border management and help anticipate and respond to breaches more proactively, said Kumar, who was earlier additional secretary at the electronics and IT ministry.
On the question of AI regulation, he said basic frameworks should be put in place to address issues of privacy, data handling, and fairness. “We can put in place a light-touch regulation, in the sense of ensuring that we do not restrict full innovations by putting in place too many expectations that they (firms) need to adhere to from day one,” Kumar said.
As India races toward an AI-powered future, getting the infrastructure, regulation, and trust equation right will be key to take it to all.
(With inputs from Annapurna Roy)
Images are for reference only.Images and contents gathered automatic from google or 3rd party sources.All rights on the images and contents are with their legal original owners.
Aggregated From –
Comments are closed.