AI's jobs apocalypse hasn't arrived yet, but the real alarm is at the entry level: the jobs that aren't being created

By ABC · October 12, 2026.

The feared AI jobs apocalypse has been postponed, at least for now. According to a recent article in The Economist cited by ABC, the boom in the technology may have helped create close to one million jobs in the United States, compared with around 200,000 layoffs attributed to AI since mid-2023.

Official data point the same way. The US Bureau of Labor Statistics recorded 162,000 jobs added in August and an unemployment rate of 4.1%.

The experts cited by the paper agree on one explanation: the ability to automate tasks was mistaken for the elimination of jobs. According to Javier Cortés of LIFT Asset Management, most professions mix tasks that can be automated with others that require judgment, accountability and client relationships. On top of that, integrating systems, redesigning processes and training teams is much slower than expected.

Goldman Sachs' own case helps put this in perspective. Its 2023 estimate (the equivalent of 300 million full-time jobs exposed to automation) measured exposure, not destruction, and the bank already warned that AI would probably complement more than it would replace.

There are important caveats to the “one million” figure. Francisco Rodríguez of Funcas warns that no one can pin down how many jobs AI creates, because official statistics do not separate jobs created by the technology from those driven by the economic cycle. The most sought-after profiles (machine learning engineers, data scientists, cybersecurity specialists) number in the tens of thousands, not the hundreds of thousands.

In Spain, according to a Randstad report, 36.1% of companies already use AI to automate processes, although adoption is concentrated among large firms.

The warning sign lies elsewhere. Stanford researchers estimate that among Americans aged 22 to 25 in the most exposed occupations, employment is around 19% below the trajectory it would have followed relative to young people in less exposed occupations. This is a relative gap between groups, not 19% of jobs destroyed, and no comparable gap appears among experienced workers.

Rodríguez sums it up this way: when substitution comes, I doubt it will take the form of mass layoffs; it will come as hires that never happen. A company that used to bring on five young people might hire only two, and that would not show up in the unemployment rate.

**Our take.** The “postponed apocalypse” thesis is correct, but it should not be read as a “false alarm.” Layoffs are the wrong metric: they are visible, political and easy to count, while the erosion of the base of the pyramid is silent. An indicator that only looks at aggregate unemployment can show good health while the entry door closes on an entire generation.

It is the same pattern we see in other sectors. Back-office and document-based tasks are the most exposed; judgment, relationships and in-person work hold up. If the law firm, the consultancy or the research department hires fewer juniors, they break the ladder on which future seniors are trained. Ana Villalba of The Valley Business & Tech School warns about this clearly: breaking that learning ladder toward senior roles is risky.

There is also a nuance that favors the optimistic thesis. Cortés argues that if AI makes legal, tax or accounting services cheaper for currently underserved SMEs, it could generate more activity and employment. It is a reasonable hypothesis, but not yet proven: it depends on productivity translating into lower prices and more clients, and on adoption that remains limited in Spain.

In the short term, the most plausible risk is not mass unemployment but polarization. Alberto Gavilán of Adecco and Luis Garvia of IEB agree that, without reforms, public support and reskilling, the technology will act as a catalyst for inequality between those who get on board with the tools and those left behind.

In the long term, our position is unchanged: an economy with mature AI points to more productivity, more abundance and people devoted to what truly adds value. But that horizon is not automatic. The transition will come in waves, and the work of the coming years is to make sure the entry door does not close while we get there.