In a recent article, ‘On building public AI, governments must be tactical but remain flexible’ (IE, February 16), Akash Kapur and Arvind Narayanan offer an important warning: Governments building public AI should resist technological vanity. Focus on enabling infrastructure, governance, and the conditions that allow innovation to serve citizens. Don’t chase frontier models. Don’t build data centres for prestige. Be tactical, be flexible.
Their argument stops one step short of the harder question. If public AI demands constant judgement about what to build, what to enable, and where markets will fail, the real challenge is not merely design. It is leadership. India has faced this moment before. When Aadhaar was conceived, the breakthrough did not come from software alone. It came from stewardship. Most visibly, Nandan Nilekani — he could align ministries, reassure industry, answer critics in Parliament and in the press, and keep execution moving at a scale that touched over a billion people. Aadhaar was not inevitable. It happened because someone with credibility across worlds chose to step into the most uncomfortable space in Indian governance — the space between a bold idea and a working system.
Transformations like that require people who turn ambition into architecture. Who convert political will into platforms. India found such leadership once. In the AI era, we need to find it again.
Kapur and Narayanan rightly point out that governments should invest in “lighter, lower-cost forms of infrastructure” for AI, things like open-source translation modules, speech-to-text systems for low-resource languages, and high-quality datasets curated for Indian conditions. They caution, correctly, that simply throwing money at large language models or domestic compute capacity is unlikely to yield public value. These are sensible recommendations. But who makes the call on which of these investments to prioritise? Who arbitrates when the Ministry of Electronics and Information Technology wants one thing, the Ministry of Health wants another, and NITI Aayog has its own roadmap? Who ensures that a promising pilot in agricultural AI in Madhya Pradesh doesn’t die because no one is coordinating with the team building a similar tool in Telangana?
The answer is not a committee. It is a person. That person must bring a specific and rare combination of qualities. Technical depth, obviously, but that is table stakes. Beyond it, they need the ability to work across governments and states in a federal system that does not naturally cooperate. They need credibility with global technology firms and with India’s own start-up ecosystem. They need seriousness about privacy and digital rights, not as an afterthought but as a design principle. They need ease with entrepreneurs and patience with bureaucrats. And they need the resilience to operate under permanent scrutiny, because any public AI programme in India will attract criticism from several directions. Above all, they need a conviction that technology succeeds only when it includes everyone — not just English speakers, not just smartphone owners, not just the urban middle class.
AI is already shaping education, health, agriculture, skilling, and welfare. These are the domains where the state touches citizens most directly, and where the gap between good design and bad design is not abstract. If the design is weak, exclusion will scale. If the design is strong, the opportunity will scale. We have seen both outcomes in India’s recent digital history. UPI scaled opportunity. The initial implementation of Aadhaar-linked welfare, before corrections were made, showed how exclusion can scale, too. AI will be no different. The difference is that AI moves faster, and the window for getting the architecture right is shorter.
Leaving that outcome to drift, hoping that the right frameworks will emerge from inter-ministerial coordination and consultants’ reports, would be a historic mistake.
For the right person, this is rare nation-building territory. The chance to lay the rails on which thousands of innovators, in government and outside it, can run. To create public infrastructure with decades of useful life. To demonstrate to the world that AI can power development, not just consumption. India’s bench of talent is deep. In technology, in public policy, in the intersection of the two, we have people who have done extraordinary things. What we need now is the courage to identify mission leadership early, empower it with genuine authority, and then stay the course even when the inevitable controversies arrive.
Because platforms do not build themselves. And India’s AI moment will not wait for us to figure out the org chart.
Thakar is a development sector mentor & director, ex-banker. Vishwanath is co-founder of MakerGhat and Inspirit and visiting research scholar at Stanford University
