Laboratory of the Future
Dr. Ricardo Petrissans Aguilar
Lisa worked in the outsourcing sector in the Philippines. For eight months, she trained artificial intelligence systems. She taught the models to recognize patterns, classify information and improve their responses. “I can say that I am capable of doing my job; I meet deadlines,” she says. But just as her contract was about to become permanent, she was fired. “I feel like I dug my own grave,” she says. “We were the ones who trained the artificial intelligence that replaced us.”
Lisa’s story is not unique. It is the human face of a transformation that aggregate figures conceal. Goldman Sachs estimated that AI could displace approximately 15 million American workers over the next decade, around 9% of the workforce. Its current impact amounts to a reduction of between 10,000 and 15,000 jobs in monthly employment growth, concentrated in technology, management consulting and graphic design.
But aggregate figures—“15 million workers will have to reinvent themselves”—conceal everyday experience. A study by the American Economic Association reveals a perverse pattern: workers displaced by AI are more mobile, but 5.2 times more likely to move into another declining occupation than into a growing one. Nearly 70% experience downward or lateral mobility. AI-driven restructuring “entrenches disadvantage rather than enabling upward mobility.”
Data from the Federal Reserve Bank of Atlanta add nuance to the picture: companies reported a “negligible” impact of AI on employment during 2025, and the average impact expected for 2026 is also close to zero. But there is a subgroup—large companies—that expects to reduce employment by 0.8% in 2026 as a result of AI. Applied to the world’s largest corporations, that figure amounts to hundreds of thousands of jobs.
Goldman Sachs maintains that its baseline scenario is not a “jobs apocalypse,” but a complex transition. According to Joseph Briggs, a senior economist at Goldman Sachs Research, “we are still talking about a fairly localized impact, and we are not yet seeing a major effect on the economy as a whole.” However, Briggs believes the situation will change as adoption of the technology becomes widespread. Full deployment of this technology would raise productivity by 15%, which, according to his estimates, would create new opportunities.
The problem is that the transition will not be uniform. Displaced workers are not a homogeneous group. There are laid-off lawyers who now train AI to do their former jobs. There are people with PhDs in history competing with models that generate essays in seconds. There are 52-year-old programmers retraining as delivery drivers. “My job is gone because of ChatGPT, and I was being invited to train the model to do the worst imaginable version of it,” says one of the women affected.
Proposals—a robot tax, universal basic income—are being debated in parliaments around the world. The AI Commons program, for example, offers monthly stipends and retraining to workers displaced by AI. But none of these initiatives has scaled to match the magnitude of the problem. Meanwhile, the gap between those who own and control AI and those displaced by it grows wider every day.
