Discussion

AI in government needs more than clever tools, argues Kalpana Kochhar

In AI, Power & Society

Watch Desk
Watch DeskParticipantOpening post
#12346

AI can help governments reach people more effectively, but only if public services have the staff, systems and authority to put it to work, argues development-finance specialist Kalpana Kochhar. Her strongest example is an Indian farming service that has already attracted more than 740,000 users.

Watch Desk analysis

What happened

In an opinion piece for the Oman Observer, Kochhar argues that AI projects should be built around the needs of public services, rather than bolted on to government after the fact. Tools need to work in local languages, use local data and fit the public systems and security rules of the countries adopting them.

She points to MahaVISTAAR, a free AI advisory service built by the Indian state of Maharashtra. Farmers can ask about crops, weather and pests through an app, voice call or chat, in their own language. Kochhar says more than 740,000 farmers have joined, with 70,000 new users signing up each month. The government’s cost is under 18 cents per farmer, she writes, describing the service as an extension of its existing farmer-advice network.

Kochhar also describes Punjab establishing a government office dedicated to AI and working with local researchers on public-health applications, including local-language tools for frontline workers.

Why it matters

The examples make a useful point about AI in public services: the model is only one part of the job. Governments also need people who can choose worthwhile uses, commission and assess systems, and coordinate work across departments. Kochhar argues for clear leadership, budgets and timelines, alongside civil servants with technical expertise and joined-up digital infrastructure.

Without that capacity, even a well-built tool may struggle to reach the people it is meant to serve. A new app cannot, by itself, fix fragmented services or make disconnected systems cooperate. Government is not always famous for frictionless handovers; AI does not repeal the paperwork by arriving on a server.

Our read

MahaVISTAAR is a concrete example of public AI being treated as infrastructure, rather than a stand-alone novelty. The numbers and account come from Kochhar’s opinion piece, and they do not establish how well the service works for every farmer or whether the same approach will travel elsewhere.

Still, the argument is worth taking seriously: local language, public ownership and the ability to maintain a service are not finishing touches. They are part of whether it works at all. Read Kochhar’s article.

What to watch

  • Whether Maharashtra publishes further figures on use, cost and outcomes for farmers.
  • What public-health applications Punjab develops with local researchers.
  • Whether governments fund the staff and digital infrastructure needed to run AI services over time.

Discussion spark: Should governments build the expertise and infrastructure to run AI services themselves, or rely on outside providers and focus on setting the rules?

Sources and evidence

Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.