Leiden, Netherlands, 6 June 2025 – SkiLMeeT researchers Najada Fejmi (LISER) and Roy Peijen (TNO) showed how information extracted from online job ads and LinkedIn data can help us understand changing skill needs, during the SkillsPULSE research seminar on June 5, 2025, at the University of Łódź.
The event, “Skills signals: leveraging online job vacancies and big data to identify skills shortages and gaps,” was organised by the SkillsPULSE project and provided a platform for three other Horizon Europe projects – SkiLMeeT, SKILLAB, and TRAILS – to present their approaches to data and discuss them with stakeholders.
Najada Fejmi presented joint research by a team of LISER researchers in her talk “Skill Demand in the Age of AI: Evidence from Europe.” The researchers used job ads data from France, Germany, Belgium, and Luxembourg (2018–2024) to explore how AI is shaping employers’ demand for skills over time. Using the large language model (Mixtral 8x7B), they extracted and classified skills from job descriptions into several categories. This allowed them to track the evolution of demand for different types of skills over time. Preliminary findings suggest that exposure to AI is reshaping what employers value in the workforce. The most significant changes in skill demand occur between occupations rather than within them. There is a growing emphasis on complementary skills, such as data analysis, decision-making, and social or communication skills, rather than on narrow specialisations. Different jobs now require a broader mix of skills than before, but within each job, AI doesn’t appear to influence how focused or varied the required skills are.
Roy Peijen (TNO) presented an analysis of using LinkedIn data to identify skills and occupational transitions. He showcased how big data can map career paths in green and digital jobs across EU countries and guide policymakers in focusing on skill-based training for effective job transitions. Peijen also addressed limitations, such as the underrepresentation of low-skilled workers and incomplete profile data, proposing solutions such as weighting techniques. His presentation highlighted the potential of LinkedIn data to inform and enhance workforce development strategies by providing actionable insights for regional policies and skill development.
