On 13 June, nearly 200 economists and AI researchers, including a dozen Nobel laureates and the chief economists of OpenAI and Anthropic, signed a blunt statement. Titled ‘We Must Act Now,’ it warns that AI “may become radically more powerful over the next 10 years,” thus transforming the economy more powerfully than the Industrial Revolution, but far faster. The explicit risk: large-scale job displacement.
What makes it remarkable is who signed it. Among them are Daron Acemoglu and Simon Johnson, MIT winners of the 2024 Nobel prize who spent years resisting such alarms. The lead coordinator of last month’s statement, Stanford’s Erik Brynjolfsson, fears the world is unprepared.
That should concern India more than most. We have the world’s largest young workforce and a formal labour market that absorbs millions of graduates yearly. If AI is rewriting who gets hired, India cannot afford to be the last to know. Our data already shows it.
The India story: Brynjolfsson’s warning follows a landmark paper with Bharat Chandar and Ruyu Chen that found US workers aged 22 to 25 in the most AI-exposed occupations saw employment fall roughly 16% over the study period versus less-exposed peers, even controlling for firm-level shocks. The count of older workers in the same offices kept growing. The paper’s authors called these young workers “canaries in the coal mine.”
In one of the first such analyses for India, we ask whether these canaries have begun to sing here. We examined monthly Employees’ Provident Fund Organisation (EPFO) payroll data, India’s most comprehensive record of formal employment, from April 2020 to June 2025.
Using the EPFO’s own categories, we grouped industries by AI exposure: high (expert services, finance and computing); medium (engineering, hospitals, schools and trading); and low (construction, textiles, hotels and cleaning). We then compared indexed employment before and after the release of OpenAI’s ChatGPT in November 2022, thereby mirroring the Stanford approach.
The results are striking. Older workers in AI-exposed sectors are surging. Younger ones are barely keeping pace.
In high-exposure Indian industries, workers aged 29 to 35 saw net payroll additions rise nearly 70% after ChatGPT’s release and 54% for those over 35. But workers under 21 barely moved: 3.4% for under-18s and a negligible 1.1% for the 18 to 21 cohort. The prime fresh-graduate age bracket of 22 to 25 years managed just 9.3%.
The contrast sharpens across categories: in low-exposure industries such as construction and textiles, even under-18s saw a 110% jump. The divergence from this is sharpest where AI has begun reshaping tasks: codifiable, white-collar work in expert services and finance.
Why might this be happening?:The global frame of analysis fits India. Generative AI, the paper’s authors argue, substitutes for codifiable knowledge: the structured, learnable work behind entry-level jobs in software, finance and services.
What AI cannot yet replicate is tacit knowledge, judgement built over years of context and experience. Older workers hold more of it; younger ones, with textbook knowledge but little context, are more exposed to AI replacement. A second mechanism compounds this: AI leverage lets experienced staff widen their span of control. Firms hire fewer entrants rather than fire mid-career staff.
Indian twist—formal-sector growth complicates the picture: A critical caveat separates India from the US: where the Stanford team finds outright employment declines for young workers, India’s formal sector is still growing.
Net additions stay positive even for the young in high-exposure industries, although they are far smaller than for older peers or the young in low-exposure ones. But the relative disadvantage is real, and where every point of formal employment matters for social mobility, it compounds fast.
The finding challenges a lazy assumption: that AI will hit India’s low-skill workers hardest. Our data suggests the opposite, for now: the young professional in expert services or finance faces the harshest headwinds.
What last month’s global statement means for Indian policy: The open letter does not prescribe detailed policy; and deliberately so. A short statement backed by 200 economists is harder to dismiss than a blueprint inviting counter-argument. It calls for incentives, guardrails and institutions that ensure AI complements rather than replaces humans. For India, three conclusions can be made.
First, higher education must urgently rethink ‘entry-level readiness.’ A degree offering only codifiable and hence easily automated knowledge is one AI is steadily devaluing. The premium now lies in applied judgement, communication and an ability to work with AI rather than be displaced by it.
Second, EPFO data (one of India’s richest, most underused labour-market datasets) deserves more attention and better access for researchers.
Third, an AI-driven skew towards experienced workers is not inevitable. Firms that invest in apprenticeships, mentorship and onboarding juniors will hold an edge as entry-level talent thins out.
Indian canaries are already singing: The Stanford researchers chose the title of their paper carefully. Canaries die first in a coalmine with toxic gases, but their deaths serve as a warning and do not pronounce a verdict. India’s young formal workers are not yet dying in career terms, but their birdsong can be detected in high-exposure sectors while their peers in low-exposure ones belt out double-digit growth.
We are yet to fully grasp this shift and our data has limits: broad categories, rough exposure classification and hard-to-establish causality. But the directional evidence is consistent, echoing rigorous US research that has led Nobel laureates and others to declare the time to act is now. The canary is singing. It would be wise to listen.
The author is lead economist and head of the Centre of Data for Economic Decision-making, Pahle India Foundation.