An OpenAI economist says AI won't destroy jobs. It's worth reading the fine print

By Momentum IA · June 30, 2026.

An OpenAI economist has come out to declare that artificial intelligence will not cause job losses. The news is circulating via Futunn, a Hong Kong financial platform, and arrives with no more detail than the headline. That, precisely, is a problem: because the argument matters a great deal, and because the source of the argument matters too.

Let's start with the obvious: that the in-house economist of the company that builds and sells the world's most powerful AI models concludes that those models are not going to destroy jobs is, at the very least, a position that warrants scrutiny. It does not mean he is wrong, but there is a clear structural incentive. Tech companies have spent decades hiring economists to produce analyses that, with greater or lesser rigor, conclude that their technology is good for employment. Sometimes they are right. Sometimes they are looking at the picture from a convenient distance.

The real debate is more complex. The classic pro-technology position holds that each wave of automation destroys old jobs but creates new ones, and that the historical balance has been positive. There is long-term evidence to support it. But there are two reasons why that historical analogy may falter this time. First: speed. Past transitions were measured in decades; the current one may be measured in years, leaving no time for training and labor mobility to cushion the blow. Second: breadth. Before, automation attacked repetitive physical tasks. Now it attacks cognitive, creative and communication tasks: the heart of the white-collar jobs that historically served as a refuge for those who lost industrial jobs.

Our reading at Momentum IA is that the truth is uncomfortable for both camps. The doomsayers who announce the end of work underestimate the human capacity for adaptation and the demand for new skills that rising productivity always generates. But the corporate optimists who say 'nothing's wrong' are ignoring what we already see in sectors such as financial services, law and the tech back office: the first waves of restructuring are already happening, and they are being painful for specific segments of workers. Telling a 52-year-old document manager that 'in aggregate, employment is not falling' is, at best, a statistical truth that is of no use whatsoever to his particular situation.

What does deserve credit in the optimistic position is this: the productivity that AI unlocks can, if it is well redistributed, fund transitions, shorten working hours, raise wages in what remains valuable and create forms of work that do not exist today. That is the long-term horizon we defend: not that there will be no disruption, but that the disruption can be the entry price to something structurally better. But that outcome is neither automatic nor guaranteed by the market alone; it requires active policies, continuous training and, yes, economists who tell the full truth, not just the half that suits their employer.

In the meantime, it's worth asking for the data behind the headline. Which models is the OpenAI economist using? What time horizon? Which sectors? Does he distinguish between net unemployment and sectoral displacement? Without those answers, the headline 'there will be no job losses' is not analysis: it is public relations.