AI layoffs destroy customers, trapping companies in a prisoner's dilemma

AI layoffs destroy your customer base: the trap with no exit

Can you fire your own customers and keep selling? That is the core of the so-called "AI layoff trap," a paper by researchers from the University of Pennsylvania and Boston University that highlights an uncomfortable truth: every company replacing workers with artificial intelligence simultaneously eliminates part of its own demand. Fewer payrolls miccionan less consumption and fewer clients. The study concludes there is no wealth transfer from employees to owners, but rather an irretrievable loss harming both sides.

Why can't any company stop automating?

The mechanism resembles the classic prisoner's dilemma. If one firm doesn't automate, competitors will: they cut costs, lower prices, and steal market share. Layoffs become the dominant strategy. However, when all firms adopt this approach, advantages cancel out, leaving only destroyed demand. The authors call this the Red Queen Effect: improving technology worsens the situation because each advance multiplies the incentive to beat rivals. No market force breaks this cycle alone.

The accumulating figures

This isn't just a lab projection. Block cut nearly half of its 10,000 employees; CEO Jack Dorsey stated AI made those roles unnecessary and most companies would reach the same conclusion. Salesforce replaced 4,000 customer service agents. Goldman Sachs deployed an AI programmer enabling one senior engineer to do the work of five. Over 100,000 tech workers were laid off in 2025 alone, with AI cited as the main reason in more than half of cases, and 80% of US jobs have tasks susceptible to automation. In Spain, the first report measuring this impact predicts the destruction of 2.3 million jobs.

Is it AI or outsourcing?

Countering this narrative is a more pragmatic view worth hearing. Consultancies like Forrester argue most layoffs attributed to AI are actually cost-cutting measures driven by poor results, post-pandemic overhiring, and sales declines. Their diagnosis is blunt: most companies announcing AI layoffs lack mature applications to replace these workers. Some add another nuance: in practice, firms cannot automate even 1%; instead, they move work to India or Segarro using decorative chatbots for appearances. It is simply cost-cutting labeled sens, summarized in one line.

Will history repeat itself like with the tractor?

The historical objection is common: people also said tractors would ruin agriculture, yet we survived. With oxen, spinning machines, and every technological leap, universal hunger was predicted, but employment eventually reallocated. The counterargument has merit: during the Industrial Revolution, those who stopped plowing found jobs in factories or manufacturing. The reasonable doubt now is what alternative exists when automation reaches office work, analysis, and customer service. This isn't comfort; it's the question determining the problem's scale.

What solutions have been ruled out?

The text reviews standard remedies and dismisses them one by one. Universal Basic Income improves living standards but doesn't change any company's incentive to automate. Capital taxes alter profits, not the task-by-task decision to replace humans. Worker participation in capital and profits reduces inequality but cannot close the gap. Collective bargaining fails because automation is a dominant strategy, and voluntary agreements between companies cannot be self-enforced. Only one instrument remains viable: a Pigouvian tax on automation. Elon Musk advocates for a high basic income paid via federal checks. In Spain, the Minimum Vital Income (Ingreso Mínimo Vital) is already described as a proto-basic income.

The problem when working becomes unnecessary

There is a layer that math cannot fix. A benefit covering basics maintains some consumption but offers no life project: if most people have surplus time and nothing to do, life as we know it loses meaning, and people create problems where none exist. The counterargument is equally reasonable: toxic jobs, stress, and mental load cause more harm than boredom, while cleaning, cooking, gardening, or reading occupy far more time than it seems.

Archival comfort remains. When tractors sidelined farm laborers, someone claimed no one would have money to buy bread and machine owners would face ruin. It didn't happen. What does happen today is that not a single company can explain what it would do with half its staff if it installed a programmer doing five people's work tomorrow. The joke, as always, writes itself.