AI and Automation Reshape Work and Inequality: gender, race, and education in American labor market 2003-22

• In Event: New Frontiers in Inequality, Poverty, and Mobility Research

Abstract

This paper advances understanding of technology and inequality by analyzing how automation and artificial intelligence (AI) selectively reshape tasks within occupations and reconfigure inequalities in the American labor market over two decades. We test four mechanisms of occupational sorting, including skills replacement, uneven automation expansion, preexisting inequalities, and institutional interactions, to examine patterns of racial, gender, and educational distribution across occupations. Using longitudinal data from the Occupational Information Network 2003-2024 (O*NET, N = 890 occupations) and American Community Survey 2003-2022 (ACS, N = 19,944,270 persons), our results demonstrate that automation's impact on skills is heterogeneous, reshaping rather than uniformly replacing manual and interpersonal work, challenging narratives of blue-collar obsolescence and the “feeling economy” perseverance. The analysis reveals automation's dualistic effect on social inequalities. Women remain underrepresented in the least automated occupations dominated by male workers, indicating persistent gender stratification in automation-resistant work. Automation exacerbates racial segregation as technological advancement disproportionately displaces Black workers from technology-intensive roles while concentrating them in moderately automated positions. Educational stratification intensifies as workers without college degrees are displaced from highly automated middle-skill jobs, yet remain overrepresented in both moderately automated work and low-wage automation-resistant service occupations. College-educated workers increasingly dominated automation-resistant occupations since the mid-2010s. By integrating task-level analysis with intersectional examination of race, gender, and education, this study offers a nuanced understanding of how automation reconfigures labor market stratification in the contemporary U.S. economy.


Keywords: automation, artificial intelligence (AI), occupational inequality, occupational segregation, race, gender, education, skills, tasks, labor market

 

Authors

• Xiaoling Shu, University of California-Davis

• Meiying Li, Cornell University

• Haowen Zheng, Cornell University