Employability in the Era of AI: Hiring for Judgment

There is one statistic that should be posted on the door of every university department, and it is not the one featured on the evening news.

In 2025, a Stanford team led by Erik Brynjolfsson—likely the economist who knows the most about the impact of technology on labor—did something unusual: instead of surveying, they looked at payrolls. Specifically, those of ADP, the largest payroll processor in the United States. Millions of workers, over seven hundred occupations, one in six employees in the country.

What they found was not job destruction. Aggregate employment, in fact, was still growing. What they found was a scissor effect.

Since late 2022, workers aged 22 to 25 in the occupations most exposed to artificial intelligence—customer service, accounting, software development—have lost 13% in relative employment. Meanwhile, their colleagues aged 35 to 40, in the same roles and within the same companies, have grown in number.

[GRAPHIC: g5-tijera-empleo-joven.png]

Read that again, because it is easy to overlook how unusual it is. The scissors do not cut across sectors, as in previous automations. They cut by age within the same role. Same job, same company: the twenty-five-year-old is no longer hired, while the forty-year-old is kept on.

What does the forty-year-old have that the twenty-five-year-old lacks?

What the machine knows how to copy

The researchers' response is the most important idea of this article. Young people compete in the market with what economists call codified knowledge: what is in lecture notes, manuals, and training videos. What can be written down can be learned by a machine. And what a machine learns ceases to be scarce.

The forty-year-old, however, brings something else: tacit knowledge. Knowing which client will cause problems before they do. Knowing which edit makes sense and which is just a passing whim. Knowing when a report is good and when it only appears to be. None of this is written down anywhere, and that is why, for now, no machine possesses it.

The nuance matters, and it should be presented fully: part of the 2022 drop coincides with interest rate hikes and the post-pandemic over-hiring hangover, and Stanford economists themselves debate how much is attributable to AI in those early months. But by 2024, with mass adoption underway, the pattern consolidated and accelerated, and has been replicated with other data in both the United States and the United Kingdom. The direction is clear, even if the exact allocation of blame is debated: the market value of codified knowledge is falling. Which is precisely the type of asset a recent graduate used to compete with.

What job postings are asking for now

If the story ended there, it would simply be depressing. But there is a second half to it, and it can be found in the job postings themselves.

Lightcast, which systematically analyzes millions of job postings, published a statistic in 2025 that would have sounded like a joke ten years ago: of the ten most requested skills across all sectors, eight are no longer technical. Communication. Critical thinking. Collaboration. Creativity. Leadership. Fully 76% of job postings request at least one; nearly half demand three or more.

[GRAPHIC: g6-ofertas-competencias.png]

The World Economic Forum confirms this from the employers' perspective: seven out of ten consider analytical thinking to be the core competency of their workforce, ahead of any technical skill. And it estimates that 39% of core skills will have changed by 2030: what you know how to do today will expire; your ability to judge will not.

[GRAPHIC: g7-pensamiento-analitico.png]

There is a third finding, and it is the most counterintuitive of all. Between 2024 and 2025, the demand for AI-specific skills in recruitment processes fell from 52% to 38%. Not because AI matters less, but because knowing how to use it is now taken for granted, just as knowing how to use email once was. What distinguishes one candidate from another is no longer that they know how to use the tool. It is what they think to do with it.

And when hires go wrong, it is almost never due to technical reasons: according to the Society for Human Resource Management, 89% of failed hires fail due to human shortcomings—communication, judgment, fit—not a lack of knowledge. In fact, 78% of hiring managers admit to having hired someone technically brilliant who failed in every other aspect.

And what about here?

In Spain, the trend lags behind but points in the same direction. Telefónica has stated institutionally that tech companies will complement technical profiles with humanities backgrounds to advance in robotization and artificial intelligence, and they are already seeing demand for philosophers and linguists in areas such as digital security and language processing. Hybrid profiles—the computational linguist, the AI ethics analyst—have ceased to be conference novelties.

And yet, let us be honest with the uncomfortable truth: Arts and Humanities graduates still have the lowest labor market participation rate of any academic branch in Spain. Anyone telling you that studying philosophy guarantees employment is misleading you.

But that was never the proposition.

The false dilemma

The mistake lies in framing it as a choice: humanities or science, liberal arts or employability. The data tells a different story.

What the Stanford scissors protect is not the philologist over the engineer. It is the person with critical judgment over the person who only has lecture notes, regardless of their major. A technical degree remains the best educational investment available: it provides depth, and without depth there is nothing to judge. What has changed is that depth alone is no longer enough, because its codifiable part—which is substantial—is becoming cheap.

The profile the market is rewarding is the engineer who writes well. The doctor who knows how to explain things. The economist who can distinguish an argument from a slogan. The developer who can envision what is worth building. A degree plus judgment. Depth plus cross-disciplinary breadth.

Such a profile cannot be improvised in a weekend workshop, nor is it acquired by watching motivational videos. It is built the way it has always been built: by reading those who thought best, by engaging in debate with others, and by writing until discovering one's own voice. In other words: through the humanities. Not instead of a degree. On top of the degree.

What a university can certify

The institutional question remains, which is the one that truly matters to those who design curricula.

For decades, universities have certified codified knowledge: this is exactly what the academic transcript displays. The problem is that this is precisely what machines are making cheap. The question the market is asking universities is not whether their graduates know things—that is assumed—but whether they possess what cannot be found in lecture notes: judgment, eloquence, and the ability to connect disparate ideas.

Companies have already begun searching for this on their own, by trial and error, in interviews and group dynamics. A university that knows how to cultivate and formally accredit this—through a real curriculum with real evaluation, not just a decorative line on a diploma supplement—will not merely be adding an ornament to its degrees. It will be certifying the only part of the transcript whose value is set to rise.

Cover image generated with artificial intelligence.