Trapped Workers: Who AI Leaves Behind

← Back to the Research Library

FIG Fellows: Daniel Plaisance, Melissa Pumphrey, Aleister Montfort Ibieta

Authors: Daniel Plaisance, Melissa Pumphrey, Aleister Montfort Ibieta, Liat Krawczyk

Abstract:

The first brief in this series, Trapped Workers: A Network Analysis of Worker Mobility in the AI Economy, introduced a framework for understanding artificial intelligence-driven job displacement, addressing both which occupations are exposed and where workers can realistically go next. Drawing on 595,000 observed worker transitions between 2019 and 2026, it mapped the labor market’s natural mobility structure and showed that, with current AI capabilities, nearly two in three highly exposed occupations are “trapped,” meaning their workers’ most likely next jobs are equally threatened by AI. Workers in these occupations can and do change jobs, but they are most likely transitioning to positions that are equally vulnerable to disruption from AI.

A growing body of research finds that AI exposure is not evenly distributed across the workforce. Most analyses point to women and younger workers as facing especially strong headwinds. This is a useful starting point. But as described in our first brief on Trapped Workers, exposure tells us only whose current job is at risk; it does not tell us whether those workers have a viable occupation to move into next. This brief applies the trapped worker framework to demographic data, identifying not just how many workers are trapped but who they are. The Bipartisan Policy Center’s analysis finds that higher direct AI exposure does not necessarily mean a demographic group is more likely to be trapped. This distinction matters, as policies based only on exposure may miss the workers and communities most in need of support.