Human Capabilities

AI Made Knowledge Cheap. It Did Not Make Judgment Cheaper.

Employers say they want skills, not facts. The evidence suggests the two were never as separate as the debate assumes.

Ameer Albahouth

14 Oct 2026 • 5 min read

The skill nobody wants anymore

When the World Economic Forum surveyed more than 1,000 employers for its Future of Jobs Report 2025, nearly every skill on its list was expected to grow in importance by 2030. Only two were expected to decline. One was manual dexterity, endurance, and precision. The other was reading, writing, and mathematics.

That second entry deserves a pause. Reading, writing, and arithmetic are the foundation of almost every school system, including the ones that educated the managers answering the survey. They are how knowledge has traditionally entered a person. Employers are not saying literacy has stopped mattering. They are saying that when machines can draft, summarize, and calculate, holding and reproducing information is no longer the thing they pay for.

That is the story most people now tell about AI and work. Knowledge is becoming a commodity, and skills are becoming the currency. The data seems to agree. Look closer, though, and it describes something less comfortable: a labor market that wants judgment more than ever, while quietly removing the places where judgment used to be learned.

How fast the ground is moving

The headline figures are dramatic, and they come from several directions at once. The World Economic Forum expects 39 percent of workers' core skills to change by 2030. LinkedIn's Work Change Report puts the figure higher, estimating that 70 percent of the skills used in most jobs will change by 2030, with AI as the main catalyst.

PwC's 2026 Global AI Jobs Barometer, built on more than a billion job advertisements across 27 countries and territories, adds the sharpest detail. The skills required for the most AI-exposed jobs are changing more than twice as fast as those in the least exposed roles. New tasks that depend on empathy, judgment, and creativity are being added two and a half times faster. Jobs that require specific AI skills grew 69 percent since 2019, against 9 percent for the jobs market as a whole, and PwC estimates the average wage premium for AI skills at 62 percent.

The usual reading is simple. What you know expires faster than what you can do. A fact learned at university can now be retrieved in seconds. The ability to frame a problem, weigh a tradeoff, or persuade a skeptical board cannot. So the sensible response is to stop memorizing and start practicing.

It is a reasonable conclusion. It also skips a step.

The junior who must already think like a senior

Buried in PwC's report are two findings that sit uneasily together. Entry-level roles highly exposed to AI are seven times more likely than the least exposed junior roles to require traditionally senior skills, such as leadership and strategic thinking. At the same time, PwC reports that the number of entry-level jobs in highly AI-exposed roles has flatlined globally.

Put those side by side and a contradiction appears. Entry-level work has always been where knowledge turned into judgment. A junior analyst built hundreds of financial models before she could sense which assumption was wrong. A junior lawyer reviewed contracts until certain clauses looked suspicious on sight. Much of that work was repetitive, and much of it is exactly what AI now does well. But the repetition was not wasted effort. It was how expertise was deposited.

If AI absorbs that layer, companies gain efficiency today. What they risk losing is the pipeline that produced their senior people tomorrow. The market is asking for juniors who already think like seniors, while removing the tasks that once turned juniors into seniors.

What the confident worker actually knows

A 2025 study by researchers at Microsoft Research and Carnegie Mellon University offers a clue about why this matters. They surveyed 319 knowledge workers who use generative AI at least weekly. Workers who had more confidence in the AI applied less critical thinking to its output. Workers who had more confidence in their own abilities applied more. The researchers also observed that people often skip critical thinking when they lack the skills to inspect, improve, and guide what the AI produces. The study relies on self-reported behavior, so it measures how much thinking people felt they did, not how many errors they caught.

Even with that caveat, the finding turns the popular slogan around. The skill every employer now wants, evaluating a machine's output, depends on knowing enough to notice when the output is wrong. Someone who has never understood how a balance sheet holds together cannot tell when an AI summary of one is quietly broken. Judgment is not an alternative to knowledge. It is knowledge that has been used often enough to become instinct.

So the real shift may not be from knowledge to skills. It may be from knowledge as a product to knowledge as an instrument of verification. People will be valued less for what they can recall and more for what they can catch.

A national experiment in Saudi Arabia

Saudi Arabia is running one of the largest tests of this question anywhere. The Saudi Data and Artificial Intelligence Authority launched the One Million Saudis in AI initiative, known as SAMAI, in September 2024 with the Ministries of Education and Human Resources. By November 2025, SDAIA announced that more than one million citizens had been trained, which it described as roughly 9 percent of the working-age population. Women made up 52 percent of participants, and employees made up 70 percent. Artificial intelligence also became a formal subject in public schools starting with the 2025 to 2026 academic year.

Employers are pulling in the same direction. In the World Economic Forum's survey, 70 percent of employers in Saudi Arabia identified technological literacy as a skill on the rise.

The ambition is clear, and the uptake is real. The harder question is what comes after literacy. Knowing how to use an AI tool is spreading quickly through the Saudi workforce. Knowing a field well enough to overrule that tool is slower to build and much harder to certify. A certificate can confirm that someone completed AI fundamentals. It cannot confirm that they would recognize a flawed recommendation in procurement, medicine, or destination planning.

That is a choice no national program can make on behalf of individual organizations. A company can treat AI as a way to compress entry-level headcount. Or it can redesign early-career work so that young Saudis still do enough real thinking to develop the judgment the market says it prizes. The first choice appears in this year's budget. The second appears a decade from now, in whether there is anyone qualified to lead.

What we are really measuring

The debate between knowledge and skills was always slightly false. Skills are what knowledge looks like when it moves. A negotiator's instinct, a doctor's hunch, and an editor's sense that a paragraph is overstating its case are all accumulated knowledge that has stopped announcing itself.

What AI changes is not the value of knowing things. It changes who does the knowing, and when. Machines can now hold the facts. People still have to hold the standards against which those facts are judged.

The labor market is paying a premium for judgment today. Whether it can still find judgment in 2035 depends on a question few companies are asking out loud: if the machine now does the work through which people used to learn, what will people learn from instead?

Sources

Note on figures: All statistics are reported as published by their sources. No figures were calculated by Arabian Growth. The 9 percent working-age share is SDAIA's own estimate. PwC's job and wage figures are drawn from its analysis of job advertisements and are not independently verified by Arabian Growth.