The European Union needs a permanent system to track how artificial intelligence is changing jobs, wages and workplace organisation, participants at a Brussels workshop concluded. The call comes as warnings of major disruption continue to outpace clear evidence of economy-wide job losses.
The workshop, held in June by the European Policy Centre and the Open Markets Institute, brought together economists, civil servants and civil-society representatives to rank the most useful indicators for an EU monitoring framework. They identified four priorities: changes in junior-level vacancies, AI exposure and adoption by occupation, shifts in firm structure linked to AI, and the labour share of gross domestic product.
Participants also highlighted a data gap. Most current measures of AI adoption rely on surveys of workers and businesses, while detailed usage data is held by AI providers. They proposed ongoing researcher access to provider data so reported adoption can be compared with actual use and employment or wage outcomes.
The proposed framework would complement existing EU labour-market tools, including skills dashboards that combine employment data, online vacancies and estimates of task exposure. The workshop also argued for an EU capability comparable to the United Kingdom’s AI Economics Institute, potentially housed in the Commission’s employment directorate and linked with Cedefop and the European AI Office. Live dashboards were favoured over occasional reports.
Existing evidence shows why such monitoring would need to cover both job numbers and job quality. It also found widespread digital monitoring and algorithmic management, including automated assignment of shifts and working time. More extensive combinations of monitoring and automated management were associated with poorer working conditions.
The Commission has separately committed to creating a high-level expert group on AI’s labour-market impact. Its tasks include developing scenarios for AI’s possible long-term effect on employment and examining analytical tools for monitoring, early warning and anticipation. The forthcoming Quality Jobs Act is also expected to address workplace AI governance, while discussions continue over training, labour protections and algorithmic transparency.
The immediate policy challenge is to distinguish theoretical exposure from actual displacement. A job may contain tasks that AI can perform without the occupation disappearing, and company spending or stated expectations do not by themselves establish real-world adoption or economic gains. A harmonised system tracking vacancies, workplace use, firm structures and labour income would give policymakers a clearer basis for responding to concentrated or sector-specific changes.