AI skills surge in job ads

A job ad for a lawyer, marketer, or factory planner now increasingly asks for one extra thing: proof you can use artificial intelligence at work, not just talk about it. In Spain, job listings that explicitly demanded artificial intelligence skills were up 555% by December 2024 versus 2023, according to the 2025 “Radiografía de empleos y sectores emergentes” report cited by El Confidencial. (elconfidencial.com) This is not just a story about software companies hiring software engineers. The same report says artificial intelligence-related roles in Spain grew more than 200% in five years, with “artificial intelligence teacher” up 2,920% and “machine learning and artificial intelligence engineer” up 1,214%. (elconfidencial.com) The shift employers are making is simple: they are buying speed. A worker who can draft, summarize, search, code, translate, or analyze with artificial intelligence can often do in 1 hour what used to take several, and PwC says industries most exposed to artificial intelligence saw revenue per employee growth of 27% from 2018 to 2024, versus 9% in the least exposed industries. (pwc.com) That helps explain why companies are paying more for the skill. PwC’s 2025 Global Artificial Intelligence Jobs Barometer found that workers with artificial intelligence skills commanded an average wage premium of 56% in 2024, based on analysis of close to a billion job ads from six continents. (pwc.com) The old fear was that artificial intelligence would simply erase jobs in bulk. The newer evidence is messier: the International Labour Organization said in May 2025 that 1 in 4 workers globally is in an occupation with some exposure to generative artificial intelligence, but most jobs are more likely to be transformed than made redundant because human input is still needed. (ilo.org) Academic research is landing in a similar place. A National Bureau of Economic Research working paper revised in September 2025 found that tasks with higher artificial intelligence exposure tend to see lower labor demand, but the overall employment effect is modest because firms that adopt artificial intelligence can also grow and reallocate work to people. (nber.org) That is why the new dividing line in hiring is often not “tech worker” versus “non-tech worker.” It is closer to “person who can use artificial intelligence inside a real job” versus “person who cannot,” whether the job is in finance, law, media, operations, or education. (ilo.org) (pwc.com) The United States is seeing the same direction of travel. CompTIA said more than 275,000 active United States job postings in January 2026 referenced artificial intelligence skills, while dedicated artificial intelligence roles such as artificial intelligence engineers and architects were up 81% year over year. (comptia.org) The catch is that employers are not asking for artificial intelligence instead of domain knowledge. They are asking for both at once: the accountant who can automate spreadsheet work, the recruiter who can screen and draft faster, the designer who can generate options quickly, and the manager who knows where the machine should stop. (elconfidencial.com) (pwc.com) That is why universities and employers are suddenly treating artificial intelligence training less like a specialist elective and more like spreadsheet literacy in the 1990s. In a 2025 Banco Santander survey across 15 countries, 81% of respondents said they felt the need to keep expanding their skills, and 7 in 10 people in Spain said continuous training will be essential in working life. (elconfidencial.com) The practical reading of all this is brutal and ordinary at the same time. The workers most at risk are not necessarily the ones in jobs touched by artificial intelligence first, but the ones still trying to do 2026 work with a 2022 toolkit. (elconfidencial.com) (ilo.org)