AI-Driven Transformation of Skill Requirements: A Human–Task–Technology Perspective

This dissertation examines how skill requirements evolve in the context of rapid technological change, particularly the advancement of Generative Artificial Intelligence (GAI) and Large Language Models (LLMs). While existing research has extensively explored skill demand through job advertisements and related labour market data, current approaches often analyse technological, human, and task‑related dimensions in isolation. As a result, there remains a lack of integrated perspectives capable of exploring skill changes across the broader skill‑change ecosystem. The dissertation therefore approaches skill change at the ecosystem‑level, involving tools, tasks, humans, and institutional responses.
The geographic and occupational scope of the analysed job-advertisement data, as well as the exploratory nature of the experimental studies, should be considered when interpreting the findings and statements presented. The research focuses on data‑, analytics‑, and AI‑related occupations. In particular Data Scientist, Artificial Intelligence, Data Engineer, Data Analyst, Business Analyst, and Business Intelligence titled positions are analysed in the job ads study method. The dissertation identifies market‑required tools and skills through job advertisement analysis, and explores possible task automation through experiments evaluating GPT’s independent performance in tasks involving the identified tools. Moreover, the dissertation investigates Human–AI interaction and cognitive‑skill dimensions in GAI‑supported problem solving. In addition, the research contributes to Natural Language Processing (NLP) in the domain through sentence‑level skill identification in job ads and curriculum–skill matching approaches, including the release of a public dataset. Finally, the dissertation conducts longitudinal job advertisement analysis to examine skill evolution, inter‑skill relationships, and the effects of AI advancements on labour market demands.
The findings suggest that Generative AI currently functions primarily as an augmentation and cognitive‑support tool rather than a replacement tool. Human supervision, analytical thinking, creative thinking, and systems thinking remain critical cognitive skills in AI‑supported environments. However, sub‑skill‑level analysis indicates that different cognitive sub‑skills benefit differently from GAI usage. The findings further suggest that GAI usage extends beyond binary usage versus non‑usage patterns and instead involves varying degrees of exploitation and collaborative interaction. Longitudinal job advertisement analysis shows that AI‑titled roles are currently more strongly associated with recent LLM tools and workflow‑oriented skills, while traditional data‑related professions continue to rely heavily on foundational technical competencies. However, across data‑related professions, traditional Machine Learning, storage and programming tools remain relatively stable and central, despite observable declines in some technologies within specific professions. Skill analysis shows that Python further appears as a central hub skill across multiple skill combinations, confirming its foundational role in modern data and AI workflows. Overall, the dissertation contributes an integrative framework for analysing AI‑driven skill change by combining labour market evidence, experimental findings, and computational skill‑matching approaches. The findings demonstrate the importance of combining multiple data sources and methodological approaches to monitor and understand evolving skill requirements.