Altman Says AI Won't Kill Jobs. Try Telling That to a Bangalore Hotel Worker Facing a 60% Wage Hike

Sam Altman, the CEO of OpenAI, told an audience in India on 26 May 2026 that artificial intelligence is unlikely to cause mass job losses. The same day, The Indian Express carried a terse dispatch from Bangalore: hotel owners in the city were warning that a minimum wage revision could raise dish-and-dosa prices by as much as 60%. Both propositions will be reported as news. Only one of them is lived.

This is not a column about who is right on the AI numbers. Altman is probably right that large-scale displacement, the kind industrial revolutions once produced, is not the immediateAI story — at least not yet, for the kinds of cognitive tasks that constitute most formal employment in Organisation for Economic Co-operation and Development economies. The evidence, such as it is, trends toward transformation rather than elimination. That is a defensible read of the current data.

But defensible reads are a luxury of the sort of room where Sam Altman speaks. The Bangalore hotel worker — a server, a cook, a clerk at a reception desk — does not experience wage legislation as a theoretical contest between productivity models. She experiences it as a number on a payslip, a number that either rises or is threatened with the alternative. The 60% figure cited by the Bangalore Hotel Owners Association over the state's revised minimum wages is not an abstraction. It is a price signal that, if accurate, passes through to an actual customer ordering an actual dosa and deciding whether the restaurant is still worth the trip.

What the optimism misses is the layered nature of modern labour markets. AI may reshape white-collar cognitive work without destroying it. But for large swathes of the global workforce — in logistics, food service, hospitality, retail, courier work, manual processing — the more immediate pressure is not algorithmic displacement. It is the same pressure it has always been: bargaining power, enforceable minimums, and the degree to which the people who do the least digitally legible work are also the people least able to absorb costs.

India's service sector employed an estimated 45 million people in accommodation and food service as of the most recent available estimates. These are not people in line for OpenAI fellowships. They are people whose wages are set not by labour market clearing rates nor by marginal product of intelligence, but by the political economy of a state minimum wage board. When that board raises the floor, the incidence falls partly on employers, partly in lower margins, and partly — as the hotel owners are now signalling — in higher prices paid by the customers who can afford them.

The framing of the Altman interview is revealing in its own right. Headlines reporting that "AI won't cause mass job losses" circulate in a news environment that treats AI as primarily a premium-sector story — something that happens to firms with research budgets, to workers with graduate degrees, to consumers of generative tools. The Bangalore story, by contrast, sits in a different register: labour regulation, small-business economics, cost-of-living pressure in a city where rents have been climbing for a decade. The two stories rarely share a headline. They rarely share an audience. That is not incidental.

The structural observation is not that Altman is wrong. It is that the question of whether AI destroys jobs is the wrong question if your interest is in the workers — in Bangalore, in Manila, in Lagos, in São Paulo — who are already in a margin squeeze from other vectors. For them, the immediate displacement risk is not a language model. It is a 60% wage bill increase that no algorithm absorbs. It is a minimum wage law that is either enforced or informalised around. It is the difference between a formal economy job and a gig.

There is a version of the AI story that is genuinely optimistic: that productivity gains from artificial intelligence will, over time, raise the value of human labour in complementary roles, and that the aggregate pie grows. That is plausible. It is also, historically, the optimistic case made for every preceding wave of automation, and the outcomes have been distributional — gains concentrated, displacement concentrated, adjustment costs borne by workers with the least bargaining power and the fewest savings buffers.

Bangalore's hoteliers are not automatons. They are employers making a rational calculation about margins in a competitive market. The workers they employ are making rational calculations about whether Karnataka's revised minimum wage, nominally a protection, translates into actual take-home income or into a price spiral that prices their employer out of business. Both calculations are correct responses to the same policy signal.

So when OpenAI's CEO tells a room in India that workers should not worry about AI, a polite response might be: whose workers? The optimists' models work fine in a world where the returns to AI accrue to firms, and where minimum wages are not simultaneously rising, and where competition keeps prices low. That world exists. It is just not one that most of the 45 million people in Indian hospitality inhabit, or the hundreds of millions in comparable sectors across the Global South.

The real question is not whether AI causes mass unemployment. It is who bears the adjustment costs of every economic transition — technological or regulatory — and whether the workers absorbing those costs are the same people the cheerleading covers. On that question, the Bangalore hotel owners' warning is more instructive than the Silicon Valley forecast. The dosa is going to cost more. Someone will decide who absorbs that.