Article Gold Open Access 2025

This time it's different – Generative artificial intelligence and occupational choice

Labour Economics
Journal · Vol. 95 · Art. 102746
Abstract

We show the causal influence of the launch of generative artificial intelligence (AI) in the form of ChatGPT on the search behavior of young people for apprenticeship vacancies. To estimate the short- and medium-term effects, we use a variety of methods, including a difference-in-discontinuity approach exploiting the exogenous nature of the unanticipated launch of ChatGPT in 2022. There is a strong short- and medium-term decline in the intensity of searches for vacancies, indicating a notable reduction in the supply of young people actively seeking apprenticeships and suggesting great uncertainty among the affected cohort. Occupations with a high proportion of cognitive tasks and with high demands on language skills were particularly affected by the decline. Interestingly, the revealed preferences in the search behavior of young job seekers contrasted with previous expert assessments on the automation risks of occupations and aligned with the most recent assessments of the AI and language model exposure of occupations – before these new assessments existed. Notably, while the supply decline did not reduce the number of signed apprenticeship contracts, we find evidence of declining applicant quality, particularly for commercial employees, the most widely offered apprenticeship in Switzerland. © 2025 The Authors

Keywords

Author Keywords

Artificial intelligence technological change Occupational choice labor supply

Index Keywords

Author Affiliations
Centre for Research in Economics of Education, University of Bern, Bern, BE, Switzerland, Swiss Coordination Centre for Research in Education, Aarau, Switzerland, CESifo GmbH, Munich, Bayern, Germany
Centre for Research in Economics of Education, University of Bern, Bern, BE, Switzerland
Funding & Acknowledgements
Staatssekretariat für Bildung, Forschung und Innovation, SBFI
The authors are grateful to the Swiss Service Centre for Vocational Education and Training (SDBB) for the provision of the national apprenticeship vacancy platform data and the financial support by the Swiss State Secretariat for Education, Research and Innovation through its Leading House on the Economics of Education (ECON-VPET). This study uses confidential data from the SDBB. The data can be obtained by applying directly to the SDBB. Replication files are available upon request (daniel.goller@unibe.ch). This study uses confidential data from the Swiss Federal Statistical Office (FSO), which can be obtained by applying directly to the FSO. The authors thank the guest co-editor and four referees for helpful comments and suggestions.
References 10 References
1 Acemoglu, Daron, Skills, tasks and technologies: Implications for employment and earnings, Handbook of Labor Economics, 4, PART B, pp. 1043-1171, (2011)
2 Acemoglu, Daron, Artificial Intelligence and Jobs: Evidence from Online Vacancies, Journal of Labor Economics, 40, S1, pp. S293-S340, (2022)
3 Autor, David H., The skill content of recent technological change: An empirical exploration, Quarterly Journal of Economics, 118, 4, pp. 1279-1333, (2003)
4 Balsmeier, Benjamin, Is this time different? How digitalization influences job creation and destruction, Research Policy, 48, 8, (2019)
5 Berman, Eli, Implications of skill-biased technological change: International evidence, Quarterly Journal of Economics, 113, 4, pp. 1245-1279, (1998)
6 undefined, (2023)
7 Cattaneo, Maria Alejandra, How scary is the risk of automation? evidence from a large-scale survey experiment, Journal of Economic Behavior and Organization, 235, (2025)
8 Chen, Jiafeng, LOGS WITH ZEROS? SOME PROBLEMS AND SOLUTIONS, Quarterly Journal of Economics, 139, 2, pp. 891-936, (2024)
9 Management Science, (2025)
10 Journal of Human Resources, (2023)
Quick Actions
Full Text via DOI
Citation Metrics
2
Times Cited (Scopus)

References 10
Document Identifiers