AI-led transformation is inevitable, but its social costs are not

India’s 2026 AI Summit showcased ambition at scale: sovereign AI models, semiconductor ecosystems, GPU capacity expansion, digital infrastructure, and India’s aspiration to become a global AI hub. The optimism was palpable despite some hiccups, AI as a productivity engine, innovation driver, sovereign entrepreneurship, and geopolitical lever generated quite some buzz.

It also ended with an AI for all declaration with next year’s summit announced in Geneva. All good so far but one question was conspicuously missing from the conversation: What happens to incomes when AI raises productivity faster than it raises wages?

On the cusp of an Engels’ pause

I have argued elsewhere that we may be entering a modern version of Engels’ pause with AI, the historical period during early industrialisation when productivity surged but wages stagnated for decades before eventually catching up.

Today, the risk is not factory automation but AI-led task automation, hollowing out entry-level and mid-skill jobs before new roles emerge at scale. For evidence, one need only look at recent layoffs at Amazon or last year at TCS.

Clearly, this risk is no longer theoretical. Recent US labour-market research, often dubbed “canaries in the coal mine,” shows early declines in opportunities for younger workers in AI-exposed occupations. These are not yet catastrophic losses, but they are early signals: concentrated displacement, slower entry-level hiring, and rising precarity for new workforce entrants.

Message from Indian data

In India, the warning signs are visible in quieter data patterns. Studies on Indian manufacturing and industry repeatedly show periods where productivity grew faster than wages, alongside evidence that labour’s share of income has declined over time. Growth has occurred and poverty has fallen, but income distribution between capital and labour has shifted unevenly.

While this happened in a pre-AI era with broad-based technological change, with AI these effects may be more acute. In other words, growth does not automatically translate into broad wage gains. That gap, the pause – matters enormously in an AI transition. Which brings us to the puzzle: why is Universal Basic Income (UBI) not even part of India’s mainstream AI policy discussion?

International interest in UBI

To respond to this, we need a quick scan of the current state of global UBI evidence. Contrary to popular belief, UBI debates are not confined to Silicon Valley intellectuals or Scandinavian welfare states. Variants of basic income thinking are already emerging across the Global South. Consider Senegal. Football star Sadio Mané has funded sustained income and welfare support in his home region, alongside investments in hospitals and schools.

In Brazil, while a full UBI is not currently implemented, income redistribution has returned to centre stage under President Lula, reviving long-standing Brazilian debates on citizen income guarantees and expanding relief for lower-income households.

India’s experiments with unconditional cash transfers

India, paradoxically, has already conducted some UBI (or more precisely cash transfer) experiments in the Global South and then seems to have quietly forgotten them. These include both national and provincial cash transfer experiments. At the union level, there is the PM-KISAN which has been a regular payout to farm households without any questions asked on the end-use of funds. At the state level there are unconditional cash transfer pilots in Madhya Pradesh, for instance, implemented with SEWA and academic partners, which showed improvements in nutrition, schooling, healthcare use, and productive investments.

Research on Rythu Bandhu shows that consumption rose

Households used transfers responsibly, often to reduce debt or stabilize consumption. The 2016-17 Economic Survey even devoted a full chapter to UBI, calling it a powerful idea whose time might come.

A recent 2025 paper in the Journal of Economic Inequality by IIM Ahmedabad economists showed that with the Rythu Bandhu scheme in Telangana, cash transfers led to an increase in consumption alongside increase in borrowings on the margin hinting at one core debate about UBI in India, which is a risk of inflation.

Populism, sir?

Indeed, the Supreme Court of India recently termed the practice of giving out state benefits indiscriminately to the public without drawing any distinction between those who can afford and those who cannot, as nothing but “appeasement, which is not conducive to the economic development of the country,” as reported by LiveLaw.

Politically, the NYAY income guarantee proposal in 2019 brought minimum income support into national debate, though it has since disappeared from mainstream policy discussion perhaps because of the appeasement argument. Yet at the very moment when AI could intensify labour-market churn, the discussion has gone quiet or is receiving the apex court push back.

Is this because UBI is a “social bad” or is it because discussing UBI is a leftist anathema one should not fall prey to? Critics often assume UBI destroys work incentives, but evidence suggests reality is more nuanced.

Cash transfers that have not had an adverse impact on employment

The Alaska Permanent Fund Dividend, perhaps the world’s longest-running universal cash transfer, has shown no significant reduction in employment, though some people shifted toward part-time work, suggesting improved job choice rather than mass withdrawal.

Finland’s national basic income experiment found little employment impact but significant improvements in wellbeing and reduced stress, important outcomes in volatile labour markets. City-level guaranteed income trials in the United States, such as Stockton’s experiment, showed improved financial stability and, in some cases, even increases in full-time employment as recipients used transfers to stabilize job searches and childcare.

Canada’s historical MINCOME experiment suggests modest labor supply reductions mainly among students and new mothers -arguably socially beneficial adjustments – alongside improvements in health outcomes. Importantly for India, Iran’s 2010 near-universal cash transfer program showed no large-scale withdrawal from work, even in a middle-income context.

Proposals for an “AI Transition Scheme”

The consistent lesson is that cash transfers change choices but rarely eliminate work. So, how might an “AI Transition Scheme” look in India? A full universal income at subsistence levels would be fiscally daunting. But policy need not jump to extremes. Consider a modest design:

1. ₹2,000 per adult per month.

2. Delivered nationwide.

3. Time-bound for five to seven years.

4. Financed through subsidy rationalization, tax base expansion, and capturing rents from high-productivity AI sectors.

5. Tapering as labour markets stabilize.

Such an AI Transition Income would cost roughly 6-8% of GDP annually, based on back-of-the-envelope calculations. These figures are certainly large but not unimaginable in a fast-growing economy, especially if paired with aggressive skilling, apprenticeship expansion, and public employment in care, climate adaptation, and infrastructure. The goal is not permanent dependency, but economic shock absorption during a historic technological transition.

Conclusion

The uncomfortable question policymakers must answer is simple: What is India’s plan for workers displaced before new opportunities arrive?

If the debate waits until layoffs become widespread, income support will arrive as crisis management rather than careful design. AI summits talk about the future of technology. They should also talk about the future of livelihoods. Otherwise, India risks mastering AI while leaving too many Indians behind in the transition.

(Views are personal, and do not represent the stand of this publication.)