The Professor Who Made AI Speak Our Languages

A Quiet Revolution from Chennai

In a world where artificial intelligence often feels like the playground of billionaires, one man from the classrooms of IIT Madras has quietly rewritten the script. Mitesh Khapra, a professor who spends more time surrounded by students and data sheets than flashing cameras, has now been named among Time magazine’s 100 most influential people in AI.

And suddenly, India finds itself in the global AI spotlight—not because of another flashy startup valuation, but because of one professor’s mission to make machines listen, understand, and respond in the voices of millions who never spoke English.

The Dream Called AI4Bharat

Back in 2019, Khapra co-founded AI4Bharat with a simple but radical idea: if AI is truly to change lives, it must speak the languages of the people.

At a time when most artificial intelligence models were built for English—and occasionally Mandarin or Spanish—Indian tongues were left behind. To Silicon Valley, they were “low-resource languages.” To Khapra, they were the living, breathing heartbeat of a nation.

So he and his team set out on a journey across India. Microphones in hand, they recorded voices in crowded bazaars, quiet classrooms, government offices, and paddy fields. They didn’t just capture Hindi or Tamil; they captured accents, dialects, and the rough-hewn rhythms of everyday speech.

The result was a treasure chest: thousands of hours of speech, across nearly all 22 official Indian languages, drawn from nearly 500 districts. Not the voices of elites, but the voices of farmers, shopkeepers, homemakers, and schoolchildren.

From the Courtroom to the Farmlands

The impact has been startling.

• In Delhi, Supreme Court documents are now translated at lightning speed into regional languages.
• In villages, farmers speak to voice bots in their own mother tongues to access subsidies or lodge grievances.
• In tech corridors, startups across India lean on datasets created by Khapra’s team, often without even realizing it.

This wasn’t just research—it was revolution. It wasn’t just about code—it was about dignity.

The Man Behind the Mission

Khapra’s journey didn’t begin at IIT Madras. It began at IIT Bombay, where he completed his M.Tech. and Ph.D. in computer science. His research in natural language processing would take him to IBM Research India, where he explored machine translation and deep learning.

But academia called him back. At IIT Madras, as Associate Professor at the Wadhwani School of Data Science and AI, Khapra found not just students, but co-dreamers. Together, they built AI4Bharat, housed under the Nilekani Centre.

Recognition followed. The IBM Ph.D. Fellowship, the Microsoft Rising Star Award, the Google Faculty Research Award—each added a feather to his cap. But if you ask him, he would say his real reward is watching a farmer use his mother tongue to access digital services for the first time.

Things Most People Don’t Know About Mitesh Khapra

1. He’s an open-source purist. Unlike global labs that lock their datasets behind corporate walls, Khapra insists that AI for languages must be free, open, and accessible.
2. He avoids the limelight. Fame was never the plan. His recognition came because the work itself was impossible to ignore.
3. He built with diversity in mind. His team didn’t just record “perfect” Hindi or Tamil. They captured the voices of illiterate farmers, urban workers, and schoolkids—ensuring AI isn’t just elitist, but inclusive.
4. He is the invisible backbone of Indian startups. Many voice-tech apps owe their existence to AI4Bharat’s data, even if his name isn’t plastered on their banners.
5. His measure of success is simple: When a villager in Bihar can ask for government benefits in Bhojpuri and get an answer, he knows the model works.

The Bigger Story

Mitesh Khapra’s recognition isn’t just about him. It’s about India’s place in the future of AI. While the West debates regulation and corporations race for dominance, India’s AI revolution is unfolding in its villages and courts, led not by moguls, but by academics with a vision.

This is influence of a different kind. Influence not measured in stock prices or Twitter followers, but in voices heard.

A Future Written in Many Tongues

When machines talk to us tomorrow, they won’t just talk in English. They will greet us in Tamil, guide us in Assamese, and argue our cases in Marathi. That future will not be built in the glass towers of Silicon Valley—it will be built in classrooms in Chennai, in research labs powered by belief, and in the relentless work of people like Mitesh Khapra.

In a world obsessed with machines mimicking humans, Khapra’s quiet revolution reminds us of something more profound: AI is not truly intelligent until it can understand all of us.

Comments

comments