The Siren Song of AI: Why Markets Can't Stop the Mass Layoff Trap
A few weeks ago, a paper appeared quietly, almost in a whisper, titled The AI Layoff Trap. Its authors, Jerry Tsoukalos and Brett Falk, are not household names. Yet their 60 pages of mathematics unfold a logic so chilling that it makes one wonder if our collective future is already being written in equations none of our policy-makers seem to understand. It is a warning — not of machines becoming sentient, not of Terminator-style robots — but of a mundane, spreadsheet-driven apocalypse. A world where companies, acting entirely rationally, hollow out their own customer base until the economy itself gasps for breath. And the most terrifying insight? They will see it coming and do it anyway.
In Delhi’s corridors of power, where “AI” is now uttered with the same blind enthusiasm once reserved for bullet trains and smart cities, such warnings are treated as academic luxury. The Indian government’s obsession with narratives of technological leapfrogging, of a “New India” powered by algorithms, carefully avoids asking the question that Tsoukalos and Falk place at the center of their analysis: if every employer replaces workers with AI, who is left to buy anything? This is not Luddism. This is arithmetic.
The Monopoly Thought: A Lesson in Restraint
The economists strip their model down to a thought experiment: imagine a single monopoly company producing widgets. Everyone works there, and everyone buys the widgets. Suddenly, AI arrives — capable of doing the work of 90% of the staff. The CEO considers adopting it. The immediate advantage is obvious: costs fall, profits rise. But this CEO is not a fool. She remembers that her own laid-off workers are also her customers. If they lose their income, who will purchase widgets? In the monopoly setting, she balances the equation. She adopts AI only to the extent that it augments workers without mass destruction of demand. She exercises what the paper calls self-restraint.
This is the benign scenario. A single rational actor with a complete view of the system can internalize the social cost. It is the logic that corporate apologists offer whenever they argue that “markets will adjust” and that “new jobs will emerge.” It is also the logic almost entirely absent from how India’s employment debate is conducted. The government’s own labour force surveys, when they were not being suppressed or delayed, told a story of shrinking formal employment, rising informality, and stagnant wages long before AI became a headline. Yet the official response remains a cocktail of denial and grand promises — Skill India, Startup India, Digital India — as if the mere naming of schemes could summon a demand base out of thin air.
The Competitive Trap: When Everyone is Rational, No One is Safe
The paper’s true, dark contribution emerges when the monopoly is replaced by competition. Now there are many widget companies, each employing workers and selling to the total pool of customers. The CEO of one of these firms does the same calculation. She can slash costs by replacing her workers with AI. But when she looks at the demand side, she sees something reassuring: her own workers were only a tiny fraction of the overall market. They never bought exclusively from her company — they spread their money across all competitors. By laying them off, she loses only a sliver of total demand, a negligible amount. The benefit of automation, by contrast, is enormous. The math is ruthless. It tells her to fire.
Now project this reasoning onto every other CEO. Each one, independently, reaches precisely the same conclusion: the individual cost of retaining workers outweighs the minuscule loss in personal sales. They all automate. They all fire. And at the end, the entire workforce — the very people who constituted the market — have no wages. Demand evaporates. Every firm, even those that automated first, finds itself staring at balance sheets soaked in red ink. The invisible hand that was supposed to guide them has instead collectively drawn a noose.
This is the AI layoff trap. It is not a coordination failure born of ignorance. The CEOs are rational. They see the cliff ahead. They know the economics of what they are doing. But as Tsoukalos explains, the trap emerges because each actor’s dominant strategy — the move that benefits them regardless of what others do — is to adopt as much AI as possible. To hold back is to be undercut by competitors who automate, lose market share, and go bankrupt anyway. “No matter what you do, no matter what the other companies are doing, your best strategy is to adopt as much of this technology as possible,” he says. This is the formal structure of a prisoner’s dilemma, the very same game-theoretic mechanism that won a Nobel Prize for showing why rational individuals can produce collectively irrational outcomes.
The Siren Song of Productivity
Tsoukalos invokes Homer’s Odyssey to explain the situation. The sirens sang with such beauty and divine knowledge that sailors, mesmerized, steered their ships onto the rocks. To survive, Odysseus had himself tied to the mast, forcing his crew to ignore his own desperate commands to turn toward the music. That rope, that external restraint, was the only thing that saved them. Today’s sirens are not mythical; they wear the unassuming garb of quarterly earnings calls and stock buyback announcements. They promise unprecedented productivity, cost savings, and competitive edge. The CEOs, like the sailors, cannot resist on their own. They need to be tied to a mast.
But here, the metaphor acquires a tragic twist. In India, the government has positioned itself not as the crew holding the ropes, but as the choir amplifying the song. Ministers speak of artificial intelligence as if it were a monsoon rain that will nourish all fields equally. The narrative is relentlessly upbeat: AI will create jobs, not destroy them; it will unlock human potential; India, with its vast pool of cheap data and engineering talent, will be the “AI garage of the world.” At no point in these celebrations does anyone in power voice concern about who will consume the goods and services produced by these super-efficient, AI-optimized enterprises. The assumption, unspoken and unexamined, is that demand will somehow take care of itself — that a new class of consumers will materialize from the data centers, or that global markets will absorb everything India produces. It is a gamble of cosmic proportions, wagered with the futures of millions who have already been rendered precarious by demonetisation, a poorly designed GST, and a pandemic that the state responded to with contempt for the working poor.
What the Paper Prescribes: Tax the Replacement, Not the Worker
Tsoukalos and Falk test several popular solutions. Universal basic income (UBI), giving workers equity, retraining programs — all are found wanting in their model, because none of them directly alter the incentive to replace a human being with a line of code. The only mechanism that works, they conclude, is to make the act of full replacement costly. They propose a tax on the replacement of workers by AI. Not a tax on augmentation — using AI to enhance a worker’s capacity would be exempt. But a firm that wants to eliminate 90% of its human workforce would face a levy heavy enough to force a genuine calculation of the social cost.
They draw a parallel with carbon taxes: a market-based instrument that internalizes an externality. A company does not pay directly for the harm of every tonne of CO2 emitted, but the tax nudges its behavior toward a societally bearable level. An AI replacement tax would operate similarly. The word “tax” is, as they concede, politically toxic, and so other instruments could achieve similar ends: subsidies for retaining workers, regulatory mandates, sectoral agreements. But the core insight remains — without an external force, without the mast to which the sailors are bound, the competitive dynamic makes collective self-restraint a mathematical impossibility.
One searches Indian policy documents in vain for any such thinking. The Modi government has, instead, presided over a steady weakening of labour laws, rebranding them as “reforms” that supposedly attract investment. A tax designed to disincentivise job destruction would represent a fundamental reversal — a recognition that capital must be made to carry the weight of its own consequences. It is a demand that would likely be met with accusations of being “anti-business” or “anti-innovation.” But the academy, through cold equations, is asking whether a society can afford to let the innovators define the terms of surrender.
A Decade Hence: Inequality as Design
Asked to imagine a United States that fails to avoid the trap, Tsoukalos speaks of a “tremendous amount of wealth inequality, something that we’ve never seen before,” and political instability reminiscent of the 2008 financial crisis. Translate that to the Indian context, and the prospect is even grimmer. India enters the AI era with a labour force already segmented by caste, religion, and region, a crumbling public education system, and a state that has perfected the art of manufacturing statistical joy. The periodic labour force survey, when released, is debated less for its numbers than for the gymnastics employed to massage them into a glow of improvement. The real story — that fewer than one in ten workers have regular salaried jobs, that real wages for most have stagnated or fallen, that the quality of employment has deteriorated even when headline unemployment drops — receives minimal official attention.
If AI-driven layoffs accelerate, the first to be excised will be those already on the margins — the contract workers, the gig deliverers, the office assistants, the call-centre employees whose linguistic dexterity was once hailed as India’s competitive advantage. The IT sector, long the poster child of middle-class aspiration, is itself turning toward automation platforms that require a fraction of the human resource. The government’s response, predictably, will be to brand any critic as a pessimist, to unfurl fresh slogans, and to commission more glossy reports about the Fourth Industrial Revolution. But the layoff trap does not care for slogans. It cares only for the circular flow of income, which, once broken, leaves behind a wasteland of defunct consumers and obsolete producers.
The Real-World Test: Can Agreements Hold?
Tsoukalos dismisses any hope that CEOs will voluntarily coordinate to slow down. Imagine, he says, that Anthropic and OpenAI agree to restraint, and Microsoft and Google follow. “As soon as you walk out of that room, your incentive … is to immediately look at the gains you can get from adopting AI, now that you know that no one else will. That’s your opportunity, and that’s your duty to your shareholders.” This is not cynicism; it is a structural condition of a system that compels even the well-intentioned toward disaster. The Indian corporate landscape, dominated by tightly held conglomerates where shareholder value is often indistinguishable from promoter enrichment, is even less likely to produce genuine voluntary restraint. The Reliances, the Tatas, the Adanis — all are racing into AI-led automation, and the government’s own policy apparatus encourages them with ever more liberal terms of operation. There is no rope, no mast, no external force. Only the sirens, singing louder each quarter.
Criticisms
- The Indian government is criticized for celebrating artificial intelligence as an unalloyed good while systematically ignoring the demand-side collapse that mass automation could trigger.
- A deliberate suppression of inconvenient unemployment data is alleged, as statistical agencies are said to have been pressured to paint rosy pictures of job creation.
- The Modi government’s “Digital India” initiative is faulted for functioning as a public-relations veil behind which labour rights are steadily eroded and industries are encouraged to substitute workers with machines without any social cost accounting.
- Economists close to the government are reproached for regurgitating outdated trickle-down narratives that assume technology automatically generates enough new jobs to replace those it destroys, a claim the mathematical models decisively dismantle.
- Mainstream news organizations are indicted for treating every ministerial announcement about AI hubs and centres of excellence as evidence of progress, while refusing to interrogate the structural unemployment already visible.
- Corporate leaders are called to account for extolling automation in pursuit of shareholder returns while disregarding the long-term destruction of the very consumer base their profits depend upon.
- International technology firms operating in India are accused of exporting AI cost-cutting models that have proven socially corrosive in their home countries, with the government offering no regulatory pushback.
- The government’s labour reforms are exposed as having stripped workers of bargaining power precisely at the historical moment when that power is most needed to negotiate the terms of technological transition.
Waiting for the firms to figure it out for themselves is, as Tsoukalos says, the worst possible thing we can do. Yet that is precisely the course charted by those who rule us, mistaking the noise of the sirens for the sound of progress.
This analysis draws on an interview conducted with Jerry Tsoukalos, co-author of the paper “The AI Layoff Trap.” The full conversation is available through the original broadcast transcript.
