Opinion What do we Indians really need from AI — and other questions we never ask

AI in India has progressed without addressing the basic issue of whose priorities it serves

What is AI but the expression of this need for regulation by the highly networked entity the Western world has become, where there is more and more management to be done?

Peggy Mohan

October 30, 2025 01:39 PM IST First published on: Oct 30, 2025 at 01:39 PM IST

An opinion piece titled ‘Mothering AI: The Adivasi Way’ by Abhik Bhattacharya (IE, October 28) set off a memory of a discussion I recently had with a techie visiting from the US. He had asked me whether I was concerned about AI and the threat it posed to humans in the long term.

I have been trying to look at AI in the context of our language universe in India. If we consider the progression of language in India, what we find is a series of layers, beginning way back in the mists of time, when we started out as small self-contained human settlements with external linkages to our neighbours. As powerful migrant groups came in, seeking lebensraum, or more space to grow and develop, they settled and interbred with local women, creating hybrid people who, in time, began speaking hybrid languages, with features from both parental streams. At each step of the way this hybridisation went hand in hand with an enlarging of terrain, with what were earlier separate groups coming together to make larger communities. At each stage of the process new needs arose: Larger human formations needed more management, which brought in kings, governments, and administration.

If this process had gone on as it started, the formation of a large nation would have brought, with a larger community, a language that identified it. Or it could have stopped, and we would have ended up as different nations identified with different languages. But that is not how it happened. Our most recent colonisers, the British, found it expedient to install their own language, English, as the language of government and administration, creating a unity at the elite level, ruling over a linguistically diverse hinterland.

Also Read | India and EU must forge a democratic tech coalition but this demands honest negotiation

But even this partial unity had its management demands, and as it linguistically linked up with a larger English-speaking universe outside, it was easy to conflate the needs of India with the systems of regulation developed in the English-speaking West.

What is AI but the expression of this need for regulation by the highly networked entity the Western world has become, where there is more and more management to be done? Western AI fills this space, having evolved from its beginnings in the 1930s from the British mathematician Alan Turing’s desire to make a “machine” that would ultimately be indistinguishable from humans in how it was able to learn by itself and “think”. Remember the Turing Test, where a human is asked to judge whether he/she is interacting with a computer or another human? We need to take this initial premise seriously: Western AI did not simply emerge as a tool to regulate a community that had grown beyond its ability to regulate itself. It started with a mission to replicate human intelligence, and then to use its much greater access to data to surpass it. That mission has not changed.

Should one be concerned about being surpassed? My answer to the techie was: “No”. Early in my days as a mother I had to confront the issue of whether I was comfortable being “surpassed”. At first my natural competitiveness, my natural wish to be the best, resisted it. But then something else kicked in. My child was not my competitor: She was my successor, and my job was to equip her as best I could for a future world where I would not be present. It did not matter if she surpassed me, because it wasn’t about “me”, in the larger context.

But what if she is able to harm you? I was asked. I stuck to my guns and said that if that was what was on the cards, it would be okay. Again, it wasn’t about “me”.

In a simple scenario, where I am the mother and AI is my child, that would work. But am I really the “mother” of a language model that originated beyond my little universe? Am I the one “training” the model? Has the existence of English in India, and its easy link to the Western world with its own priorities, produced an illusion of relationship to this “child”?

And that is the crux of my concern. AI in India has progressed without addressing the basic issue of whose priorities it serves. The more I look the more I see bright-eyed grinning techies, the successors of the British-era ICS officers, keeping their heads down and working. Proceeding with details without the most basic questions being decided. Who are we? What do we need done by AI? And, most important of all, is there only one AI?

The appearance of the Chinese chatbot DeepSeek, trained on very different models (which are cheaper and more efficient in terms of utilization of computer resources) brings this question into sharp focus. We have been negligent in taking the easy way out and tagging on to Western AI and hoping for a free ride. There is no free ride.

Can techies sort this out? That would make as much sense as wondering if individual fleas on a dog could change its direction. Late in the day we have to start wondering if we do have our own place in the world, and a future that is not dependent on a West that seems to be going into decline.

What would our own AI look like?

The writer has taught Linguistics at Howard University, JNU and Ashoka University. She is the author, most recently, of Father Tongue, Motherland: The Birth of Languages in South Asia