India must put policy first, politics later to realise its AI Ambition

The country must institutionalise the technology with the sobriety that once anchored Aadhaar and implementation of GST

Leaders, dazzled by coalition arithmetic or the sugar rush of a front-page headline, slap slogans on problems they haven’t bothered to diagnose. Think Brexit: A plebiscite staged for short-term partisan gain, no road map in the drawer, and which had the entire British nation twirling. Or America’s 1971 wage-and-price freeze: Televised bravado that spiked Richard Nixon’s poll numbers, only to unleash shortages and usher in the era of stagflation. When the cart is put ahead of the horse, the journey often ends in a cul de sac.

 

There is, fortunately, a saner sequence. History is littered with moments when hard policy legwork quietly laid the tracks before the political gravy train could glide in. West Germany’s 1948 currency overhaul punctuated hyper-inflation and birthed the Wirtschaftswunder; citizens soon rewarded the architects with a deep well of legitimacy. Brazil’s Plano Real, mid-1990s: inflation slain, public relief palpable, and economist-turned-steward Fernando Henrique Cardoso swept into high office on that gust of credibility. Botswana’s bold decision to foot the bill for anti-HIV drugs? AIDS deaths collapsed by roughly three-quarters, and the government’s reputation blossomed.

 

So then why do so many leaders still sprint toward the press podium before they’ve ironed out policy kinks? A few culprits leap out. Election cycles, for one, click by like stopwatch seconds; tomorrow’s vote feels tangible, whereas a ten-year reform dividend might as well be on Mars. Coalition politics doesn’t help — bold measures upset delicate partner egos, so timidity masquerades as pragmatism. Add human frailty: present-bias, face-saving, the gambler’s overconfidence that charisma can outrun consequence. Small wonder the urgent displaces the important.

 

India knows this truth, at least on its better days.

 

The country has had its tryst with putting politics before policy. However, there have also been instances where India got the choreography right. Consider Aadhaar: It didn’t burst forth with drum beating slogans; instead, it began as an engineering quest — untangling the knot of identity verification, privacy, and brute force scalability for 1.4 billion Indians. Biometric glitches and data security potholes were fixed, one bug at a time. Slowly, the platform proved its mettle in everything from bank account seeding to leak proof subsidies. Only after that quiet slog did Aadhaar stride onto centrestage, lionised across the aisle as the emblem of welfare modernity.

 

The GST story was not different. Long before the thunderclap of “one nation, one tax,” the reform was built in backrooms with the Centre and states squabbling, compromising, then squabbling again. Dozens of marathon sittings forged consensus, showing that cooperative federalism can double as a forge rather than a speed breaker. Three decades earlier, Manmohan Singh’s liberalisation-privatisation-globalisation policy loosened the licence raj straightjacket first, letting the GDP roar. Campaign leaflets trailed behind, gasping for relevance. The moral? Robust design, open consultation, and institution building conjure the fuel for political showmanship later. Voters, it seems, possess an uncanny ability to sniff the gap between fireworks and foundations.

 

Overcoming India’s AI frontier

 

India now stands before an unruly frontier: artificial intelligence. The assets are enviable — coders by the million, a skittering startup bazaar, and robust digital public infrastructure that has become, dare one say, export quality. Yet, the challenge is formidable. India’s planned public private high end GPU clusters (about 18,000) only begin to approach the scale on which today’s frontier models are trained. OpenAI’s early ChatGPT versions used roughly 25,000 GPUs (graphics processing unit). Securing next generation chips will remain a moving target, made harder by geo strategic export controls and India’s still nascent semiconductor fabs.

 

To win this time, we will need to repeat the script — policy before politics — with a stronger resolve. We should start with work. AI is rewriting the terms of employment faster than one can spell transition. India, staring at 60 million entrants to the labour market this very decade, cannot afford to improvise. TCS’s decision to shed 12,000 positions is a tremor, not an earthquake. Governments must absorb the shock upfront: Budget for mass reskilling, wire social insurance floors beneath informal workers, and embed adaptive curricula in every skilling mission. Equally important are social safety nets and labour reforms to cushion displaced workers during transitions. India’s initiatives like the Skill India Mission (2015) and the Pradhan Mantri Kaushal Vikas Yojana are steps in the right direction that will need further strengthening.

 

Next comes power. Semiconductors are the new coal, and Beijing and Washington are already digging. China’s third “Big Fund” stands at $47.5 billion; the United States’ CHIPS Act disburses $52 billion and unlocks another $200 billion for science. India’s ₹10,300 crore IndiaAI Mission is a necessary down payment, not the endgame. Without stable incentives, open data commons and predictable chip supply lines, we risk becoming tenants in an AI economy designed elsewhere. Sound industrial policy must therefore pre empt electoral pyrotechnics; otherwise, slogans will travel faster than silicon.

 

Finally, democracy itself is under algorithmic audit. During the 2024 general election, deep-fake content seeped into the public realm; many voters, unable to tell illusion from evidence, accepted them as fact. Add the silent proliferation of facial recognition cameras, and you have a cocktail of manufactured consent and ambient surveillance. A constitutional republic cannot outsource truth testing. The Digital Personal Data Protection Act of 2023 lays a foundation, but sector specific guardrails on bias, explainability, and citizen redress must follow swiftly. Open by design remains the gold standard: India Stack’s transparent APIs proved how scrutiny crowds in innovation, while BHASHINI extends that logic to language AI.

 

So, where do we land?

The lesson is as old as Kautilya and as ignored as yesterday’s manifesto: Politics and policy are not enemies; sequence is everything. Sound policy begets durable politics; empty sloganeering seeds cynicism. In the AI age, delayed policy is not merely bad governance; it is an existential liability. If India can institutionalise AI with the sobriety that once anchored Aadhaar and GST, it won’t merely harness a technology; it will refurbish the social contract. Legitimacy will flow precisely because leaders sweated over the plumbing before snapping the selfie.

The writer is a global policy expert and country director, Tony Blair Institute for Global Change.   These are the personal opinions of the writer. They do not necessarily reflect the views of www.business-standard.com or the Business Standard newspaper