India has emerged as a major hub for AI localization. Credit: Microsoft
- PODCAST
Episode 52: This Week in India's Science: 26 May 2025
doi: https://doi.org/10.1038/d44151-025-00095-1
Never miss an episode: Subscribe to the Nature India Podcast on Apple Podcasts, YouTube, Spotify, Acast or your favourite podcast app. Head here for the Nature India Podcast RSS feed.
Transcript
Cosmologist Jayant Narlikar’s legacy and the Global South’s push for culturally rooted AI
Host: Subhra Priyadarshini; Sound editing: Prince George
*****
Welcome back to This week in India’s science, brought to you by Nature India. I am Subhra Priyadarshini.
*****
A towering figure in India's science has passed. Jayant Vishnu Narlikar — astrophysicist, cosmologist, science fiction writer, and institution builder — died last week in Pune at the age of 86.
Best known for co-developing the Hoyle–Narlikar theory of gravitation and challenging the Big Bang model, Narlikar wasn’t afraid to stand outside the scientific mainstream if it meant pursuing intellectual clarity. His work, alongside Fred Hoyle, proposed an alternative model of the universe — one without a beginning — and sparked decades of lively debate.
But Narlikar’s influence stretched far beyond cosmology.
In 1988, he founded the Inter-University Centre for Astronomy and Astrophysics (IUCAA) in Pune, an institution that transformed access to research for students and scientists across India. He designed it as a space for both rigorous science and open collaboration — a kind of scientific democracy.
He was also a gifted communicator. One of the country’s earliest and most beloved writers of science fiction and popular science, Narlikar brought black holes and time dilation into Indian living rooms — and classrooms.
Jayant Narlikar: I think that there is reason to be optimistic when you look at the younger generation like schoolchildren. They are getting exposed to more of scientific ideas and probably they will end up more science friendly or rational thinkers than their parents have been. So far as the older generations are concerned, I sometimes feel very disheartened. The number of superstitions that they are likely to have has, in fact, gone up with time. Like weddings or marriages are fixed according to matching of horoscopes when scientifically it has been debunked any number of times.
That was Narlikar from an interview a few years back. In one of his last public reflections, he said: “The universe is not obliged to follow our prejudices. But we are obliged to understand it honestly.”
A fitting epitaph for a life lived in pursuit of truth and many times against the cosmic current.
You can read the full obituary by Pranav Sharma on Nature India.
*****
In other news this week:
While the US and China race to build ever-larger AI models, researchers across the Global South are quietly reshaping the field — in their own languages, and on their own terms.
From India to South Africa, Egypt to Saudi Arabia, new language models are being trained not just for technical performance, but for cultural relevance. In India, government-backed initiatives like BharatGen and labs like AI4Bharat are building tools for farmers, students and small businesses in over 20 Indic languages. Across Africa, collectives like Masakhane and companies like Lelapa AI are prioritising community data and open-source design.
The models are smaller, but more grounded — tuned to local dialects, scripts and meaning. It’s a bold vision of AI that speaks with the world, not just to it.For more on this emerging story, head to Nature India, Nature Africa and Nature Middle East. We have a whole collection of articles this week on AI models in the global south.
Balaram Ravindran: Most large language models are trained on English content, missing the nuance of India’s linguistic and cultural diversity. That’s why many of today’s AI tools don’t serve Indian users well — especially in regional languages or rural settings. But more and more people are trying to find solutions for this challenge. And I am very hopeful.
That was Balaram Ravindran, who heads a number of AI initiatives across India. Ravindran talks about efforts that are creating rich datasets in Indian languages in a comment this week in Nature India. Startups and research labs are building smaller, efficient models that can run on smartphones — bringing AI to farmers, health workers, and students.
*****
That was this week. Stay updated with the latest from the world of science in India on this podcast. This is Subhra Priyadarshini and you were listening to This Week in India’s Science on the Nature India podcast.
