Inclusive AI and India's Digital Public Infrastructure

Over the past decade India did something that few expected and many are still studying: it built digital public infrastructure that reached across the whole of society. Identity, payments and data-sharing were reimagined as shared public rails rather than private products — open, interoperable, and designed so that the smallest player and the largest could use the same foundations. Whatever one’s view of the details, the achievement is undeniable. A billion people were brought into digital systems at a speed the world had not seen.

As artificial intelligence becomes woven into public life, it is worth asking what that experience teaches. Not the specific technology — AI is different in kind — but the spirit behind it. India’s digital public infrastructure succeeded because it was built on a particular set of instincts about inclusion, openness and the public good. Those instincts are exactly the ones that AI in the public sphere will need.

The lesson of building for the many

The defining choice in India’s infrastructure story was to design for the many rather than the few. The systems were built to work for a person with a basic phone in a small town, not only for the well-connected urban user. Inclusion was not an afterthought bolted on once the product was finished; it was the starting assumption that shaped every decision.

This is the first lesson for inclusive AI. Left to their natural course, AI systems optimise for the users who are easiest to serve — those with good data, good connectivity, and a language the system was trained on. Building for the many means deliberately resisting that pull: ensuring that the systems work in Indian languages, for Indian contexts, for people at the far end of the digital divide. Inclusion has to be a design principle from the outset, not a later act of charity.

Openness as a public value

A second instinct that made India’s infrastructure powerful was openness. By keeping the core rails open and interoperable, the country prevented any single player from owning the foundations of public digital life. This allowed a wide ecosystem to flourish on top — small innovators alongside large ones, all building on shared ground.

AI raises the same question in a sharper form, because its foundations are expensive and tend to concentrate in a few hands. An inclusive approach asks how much of the AI that touches public life can rest on open, shared foundations — open models, open datasets that respect privacy, shared standards — rather than being enclosed entirely within proprietary systems. The point is not hostility to enterprise; it is a recognition that when infrastructure is public in spirit, the benefits spread far more widely than when it is locked away.

Inclusion is a moving target

One caution the infrastructure experience offers is that inclusion is never finished. Building the rails was not the end of the work; there was, and remains, a continuing effort to reach the last person — those without documents, without devices, without confidence. The systems that reached a billion people still had to keep reaching for the ones beyond that billion.

AI will be the same, only more so, because its exclusions are subtler. A person can be formally included — able to access a service — and still be effectively excluded because the AI behind it does not understand their language, their accent, their name, or their circumstances. Inclusive AI therefore demands ongoing vigilance: continually asking who the system still fails, and treating each answer as unfinished business rather than an acceptable edge case.

Guarding against new divides

There is a real risk that AI, if approached carelessly, could undo some of what inclusive infrastructure achieved. A society that worked hard to bring everyone onto shared digital rails could find a new divide opening on top of them — between those whom AI serves well and those it serves poorly, between those who can direct these tools and those who are merely subjected to them.

Avoiding that outcome is a matter of intent. The same intent that drove the infrastructure story — a determination that the technology should serve the whole of society and not fracture it — has to be carried forward into the age of AI. This is not automatic. It requires people and institutions who hold inclusion as a genuine commitment and keep asking the uncomfortable question of who is being left behind.

Carrying the spirit forward

India’s digital public infrastructure was, at its heart, an act of faith that technology could be built as a common good — open, inclusive, and reaching the many. That faith is now being tested again by AI, a more powerful and more concentrating technology than any that came before.

At Svadrishti Foundation we believe the answer lies not in inventing new values but in carrying forward the ones that already served the country well. Build for the many. Keep the foundations open where you can. Treat inclusion as unfinished work. Guard against new divides before they harden. This is what an inclusive approach to AI looks like in the Indian context — less a novel doctrine than a continuation of the same public-spirited instinct, applied with fresh urgency to a new and formidable technology.