Raisina : Welcome To the Machine

Every transformative technology arrives bearing the same promise: that it will flatten old hierarchies, widen opportunity, and bring the bright future within reach of those who have been waiting longest for it.  And every time, arguably, the same thing happens.

The printing press empowered the literate and the wealthy. The internet enriched those wealthy enough to be early adopters and already connected. Now comes artificial intelligence, the most capital-intensive and structurally complex technology in human history. And the question on the table at Raisina Dialogue 2026 on Friday was both simple and urgent: will it be any different this time ?

The session, titled "Intelligence Bias: Equity, Inclusion and Growth in the Age of AI," convened a deliberately global panel — former Indian Union Minister Smriti Irani, US Economist Meredith Walker, University of California San Diego Dean Rajesh Gupta, former Thimphu TechPark CEO Tshering Cigay Dorji of Bhutan, and moderator Paula Cipierre, Global Head of Privacy at HCL Tech in Germany. The geographical spread was impressive.

Ms Irani struck an optimistic note, reminding the audience that AI has democratised the usage of technology. "It is incumbent upon us as citizens to use technology to a much more productive end," she said. "We have today an opportunity to address a lot of challenges that the Indian citizens and government deem fit for attention."

Such optimism is found in many parts of India, with its population-related issues and uneven development. AI-powered tools have already enabled farmers in remote Indian villages to access weather forecasting, legal aid and market prices via a basic smartphone. India’s Digital Public Infrastructure model has been described as a potential "Manhattan Project for AI" aimed specifically at expanding digital inclusion for rural populations and hundreds of millions of informal workers, most of them women.

An idealised version of farming in India created by AI

These are real gains. The question is whether they constitute inclusion, or merely the first visible edge of an AI wave that will wash over many in India, only to run up some foreign beach.

The structural concern is this: AI is not cheap to build, and it is not neutral in its construction. The large language models and foundation systems at the heart of the current revolution have been trained overwhelmingly on English-language and Western data. The biases involved are not incidental — they are baked into the architecture.

An AI hiring tool trained on historical employment data will potentially replicate historical employment discrimination. A medical AI trained on clinical data from the populations that have long dominated research trials will perform less reliably on everyone else.

There is also the compounding problem of what analysts call the qualitative gender gap. More women are digitally literate today than ever before — but digital literacy generates less economic value for women than for men, because the sectors, roles, and opportunities that AI is transforming now are disproportionately those where men are already employed.

The panel's composition itself demonstrated this issue. Small island economies like Mauritius, landlocked Himalayan kingdoms like Bhutan.

Tshering Cigay Dorji posed the question :  how does a nation of less than two million people, without its own AI infrastructure, position itself to benefit rather than be bypassed? The answer, for now, is mostly dependency, although Bhutan has moved to develop its own AI model to teach its national language Dzongkha, but still using AI tools built elsewhere, governed by rules being developed elsewhere.

That regulatory question is one of the sharpest the session confronted. Historically, regulation of transformative technologies follows implementation — often by years or decades — for example we didn’t have traffic lights until it became obvious cars were going to crash into each other at intersections.

But as Rajesh Gupta from UCSD pointed out, “We are used to having machines where we have the agency, but now we have situation where something else has the agency.”

“When humans have the agency, they operate with rules and laws, but that may not be the case [with AI], the rules may change, and that’s a very unsettling scenario.” 

The question is whether AI governance can be front-loaded: whether the rules shaping who benefits and who bears risk can be written before the technology embeds itself so much that changing course becomes politically or economically impossible.

The  Raisina Dialogue framed this challenge in its 2026 programme: Can technology bridge gaps that politics has widened, and what does gender equity mean in a digital, post-pandemic economy?

US economist and cyber-expert Meredith Walker reminded the audience of the US Constitution’s guarantees of freedom of religion and, perhaps more importantly, freedom of conscience.  “That’s what makes us fundamentally human, the ability to choose what we believe, who we associate with, and what types of beauty we want to create in this world. – that is my hope for AI” she said.

Walker is currently with the Dallas-based Cyber Future Foundation, helping Americans deal with their own AI issues, and attempting to influence the development of the technology.

From an economic perspective, she makes the point that the modern global economy is based on the ability to price risk, and when things go wrong, to be able to determine who should pay. “I do think AI raises important questions about that, who is going to be accountable – and it does need to be a human.”

“Fundamentally we need the freedom to build businesses and collect rewards, or take our losses, but something like AI interferes with that freedom to enterprise, and that’s a concern on a macro level.”

A far more accurate picture of farming in India, with women shouldering much of the hard work / Image pexels

All of these gaps and questions around AI present a difficult scenario – in which AI rollout actively builds on existing problems by concentrating productivity gains in already-productive places, automating away the many service jobs that have historically provided economic entry points for women and migrants, and giving big states and big tech ever-greater leverage over digital infrastructure.

The panel ended without a tidy resolution, because the problem doesn’t currently have one.

What it offered instead was a useful reframing. The question is not simply whether AI will include everyone - the answer is almost certainly yes, in some form.

The question is what the terms of that inclusion will be: who sets them, who enforces them, and who has a seat at the table when they are written. In New Delhi on Friday, at least, some of those voices were in the room.

Asia Media Centre