Experts at India Summit: AI Should Be Aimed at Clear Business Problems

With the advent of generative artificial intelligence, reactions to the transformative technology have alternated between fears of human displacement and unbridled techno-optimism. 

For innovators at some of India’s largest companies, AI holds vast potential to solve previously intractable problems, but the solutions are only as good as questions behind its application, experts said during the 15th annual USA India Business Summit at Georgia Tech Feb. 12. 

Large-language models, or LLMs, provide the underpinning for platforms like ChatGPT or Claud. Inhaling troves of text from across the Internet, these models train computers to provide sophisticated and near-instantaneous responses to user queries. Use cases abound, from drafting essays and synthesizing research to crafting resumes and even writing code. 

But the approach comes with a cost, both to society and to individuals— in terms of computing power, skyrocketing demand for electricity needed to power data centers, and ultimately in fees passed to the user. (Among ChatGPT’s 400 million users, about 10 million are paying subscribers). 

For the enterprise, however, defining the business problem AI is aiming to solve can help with uncovering less costly and more relevant solutions, says Lakshmanan Chidambaram of Tech Mahindra. 

“We believe that you need to focus on verticalized LLMs and SLMs, small language models, that don’t boil the ocean but take a particular problem in a particular industry, and then build an LLM to solve it, or an SLM to solve it,” said Mr. Lakshmanan, president and head of the Americas Leadership Council at Tech Mahindra. 

That insight came after TechMahindra, the IT outsourcing affiliate of the Mahindra, the conglomerate making more tractors than any other company in the world, created The Indus Project, India’s first indigenous ChatGPT rival, based on Hindi language and related dialects spoken by 650 million Indians. The effort is being extended to India’s 27 other official languages, building off the Bharat markup language, also created by TechMahindra’s innovation arm, Makers Lab, to help non-English speaking coders. 

“We operate where information technology, business process, core engineering meet and networks meet,” Mr. Lakshmanan said during a session moderated by UIBS Managing Director Ani Agnihotri. “That’s the magical intersection we work in.”

Among many use cases, from educating young people to preserving dialects, Mahindra saw the potential for AI to improve the experience for India’s 1.4 million mostly small-holder farmers. 

The company created an AI “buddy” for farmers, who could send location-based queries and receive real-time responses from an bot recommending appropriate fertilizers and planting times based on weather and soil conditions.

Mahindra extended the idea to smart tractors that can use audio to inform their owners in local languages, such as Bhojpuri, that they need maintenance in a certain area. 

The solutions, once the problems are defined and now that India has its own LLMs, are potentially endless, Mr. Lakshmanan said, providing the example of a woman calling from a cell phone to get prenatal care in her own language from an AI agent. 

The magic, he said comes when LLMs combine with small-language models, those trained on a particular data set with defined parameters, and so-called edge AI, where the capabilities are embedded in applications on servers close to the user, consuming less electricity and computing power. 

“And now just imagine you tie it up with agentic AI: workflows. Ultimately, there’s somebody who’s doing particular work in a shop floor, in a hospital or elsewhere in agriculture, and as they are doing the work, there are a number of intelligent bots which are understanding what needs to get done and then pulling information together to help the human being deliver spectacular results. So it’s exciting, a combination of verticalized LLMs, SLMs, AGI and agentic workflows coming together.”

TechMahindra is already implementing this in health care but also in manufacturing for Mahindra’s own uses. In Chakan, just outside of Pune, 14,000 employees work in concert with 400 robots and 12,000 machines to churn out Mahindra SUVs and pickups that in India need to be priced at just about $25,000 to remain competitive. A new EV factory launched in January on the site will have 1,000 robots. 

AI, mostly through computer vision, is helping improve safety, boost productivity and drive sustainability on the factory floor, Mr. Lakshmanan said, citing examples in which cameras detect when employees in sensitive areas forget their hardhats and computer systems conduct root-cause analyses when a sealant is applied incorrectly or batch paint shop orders to reduce energy usage. 

Back upstream, AI is being applied to reducing costs for everything from back-office support to vehicle research and development. The 2025 Mahindra Scorpio, India’s top-selling SUV, was developed in India for about $120 million, versus $1 billion or more that legacy foreign automakers may have spent, Mr. Lakshmanan said. 

“We believe that AI is like three spoons of sugar and a glass of milk in every single part of the operation. It’s going to make everything efficient,” he said. 

Because of that, he added, leaders need to begin conversations with workers now on how to improve their productivity with artificial intelligence. 

“You can’t wish this problem away. Human beings will have concerns, and change is always difficult, so you need to work on human machine collaboration,” he said. 

On this front, Sandra Jordan, chief of staff at the Southern Association of Colleges and Schools Commission on Colleges, an accrediting body, said more emphasis should be placed on AI ethics. 

“There will be displaced jobs, there will be change, just like with the Industrial Revolution,” Dr. Jordan said in conversation with Georgia Tech CIBER’s John McIntyre and Swiss Consul General Urs Broenimann. “We can already sort of see it coming. So the question is, how do countries prepare …to retrain and retool and help their citizens move forward.”

At Atlanta-based Graphic Packaging International, the implementation of new technology is similarly driven by the pursuit of manufacturing efficiencies, said Vish M. Narendra, senior vice president and chief information officer. 

The operator of paper mills cranking out products ranging from paper cups to folding cartons and beyond will soon open a $1 billion connected factory in Waco, Texas, that makes use of insights generated from sensor-enabled digitization of physical assets at older factories. 

New initiatives, Mr. Narendra said, are easier for executives who are optimistic about AI to fund if they can be explicitly tied to the bottom line, even if the improvements are incremental, he said. 

A 1 percent improvement on a machine that produces 4 million tons of paper, for instance, results in 40,000 more tons from the same asset over the course of a year, he added. 

“If I can speed that up and produce more, that is dramatic productivity, both top-line and bottom-line growth for the company, that is a bet that we are willing to take, as opposed to saying, you know, let’s deploy Microsoft Copilot and I can write an email three seconds faster. So, it really is important to think about how you how you leverage AI within your business, and it has to be in the context of true business value outcomes.” 

He added that technology, from virtual and augmented reality to AI, is aiding with the recruitment, training and onboarding of tech-savvy employees replacing those aging out of the labor force. 

“If you told a brand new graduate that came on the job that this 300-page binder is their work, they’re going to look at you like you have two horns,” Mr. Narendra said. “These are kids that grew up on Call of Duty, grew up on Grand Theft Auto. They are digital natives, so they really need things in a digital way, and that’s the only way that they’re going to operate, consume and learn. Creating a connected worker is really important for us.”

During the session on artificial intelligence, Stan Sthanunathan, a former Coca-Cola Co. and Unilever executive and the CEO of i-Genie.ai, also shared his insights on the use of AI to help brands glean unrivaled consumer insights on from a panoply of sources including social media. 

Themed around “Sustainability, AI & Digitalization: Global Drivers of Growth,” the two-day USA India Business Summit included sessions on U.S.-India trade, Indian tech talent, AI’s impact on marketing, and keynotes and insights from executives from firms including Cantaloupe Inc., Infosys Ltd., Georgia-Pacific, UPS, Novelis and more.