This project examines the roles of labor market policies, including Unemployment Insurance and workforce development programs, in helping unemployed workers adapt to AI.
Background
Artificial Intelligence (AI) technology has rapidly advanced over the past two decades, most recently with the widespread diffusion of generative AI tools that began in late 2022. AI is expected to greatly alter labor markets, including increasing job losses and “churn” by impacted workers who switch between jobs. However, very little is known about the extent to which social insurance programs are equipped to address AI-driven displacement. Unemployment Insurance (UI) is the primary US program to help workers after they lose their jobs. UI could potentially help AI-displaced workers with financial support as they search for new jobs and adjust to the evolving future of work. If UI proves insufficient for supporting AI-displaced workers, workforce training is another option for helping workers reskill for new roles, including ones that may require AI skills.
As AI continues to impact workers and companies, policymakers also need to know if AI will impact the UI system in different ways. For example, if AI-impacted workers tend to have higher education levels, then there may be greater demand on the UI system if higher educated workers enroll in UI at higher levels than the average worker, and are higher income. Alternatively, AI workers may be impacted by ongoing technological disruptions that permanently alter the types of jobs in demand, necessitating upskilling (i.e., job training) or certification that makes workers more “AI-proof.” These shifts could mean that people are unemployed for longer periods of time and policymakers may want to extend the amount of time that people can receive UI benefits.
As the world’s fifth largest economy, California is an ideal state to study AI’s impact due to its size, data, and proximity to AI innovations.
Research Project
This research project seeks to answer these questions:
1. Do AI-exposed workers enroll in UI at similar levels as other workers, and do they receive UI benefits for similar lengths of time as non-AI impacted workers?
2. How might changes in UI benefit amounts and the potential duration of benefits affect employment and earnings outcomes for AI-displaced workers?
3. Are UI benefits an adequate (temporary) safety-net for AI-impacted workers during the time they are looking for new employment, or do these workers also need additional support, such as training and certification, in order to regain employment in AI-reliant jobs?
4. Are AI-impacted workers able to use UI benefits to better prepare themselves for this new AI reality, or is additional training and certification required for work in AI reliant jobs?
We will study how well California’s workforce development programs prepare workers for AI-intensive jobs by measuring their labor market outcomes before and after the training, including employment status, earnings, and whether the worker’s subsequent job is in an AI-using occupation or industry.
Research Team
Dr. Benjamin Hyman (Co-Principal Investigator), Professor Till von Wachter (Co-Principal Investigator), Dr. Roozbeh Moghadam and Dr. Swapnil Motghare
Results
Forthcoming