Artificial intelligence (AI) represents a seismic technological shift that has potentially transformative economic implications. With widespread adoption of AI among workers, the nature of work is changing at an unprecedented pace. And, as with past technological changes, the expected impacts are likely to be uneven. 

This brief is a contribution to the Black Family Thriving Initiative, which aims to develop insights that drive economic mobility and wealth building for Black families. As AI represents a significant shift in the labor market and could affect wages, economic mobility, and wealth building, we seek to understand how Black workers may be impacted. We provide updated evidence of the relative exposure of Black workers to AI and its potential consequences, with a focus on Black middle-class workers.

Why This Matters

Understanding the potential impacts of AI on Black workers is essential in part because of the possible role this technology will play in shaping opportunities and outcomes that are central to building the Black middle class, including not only the returns to education and work and the paths provided by the labor market to economic mobility but also the importance of building wealth for economic security. Early research on the potential effects of AI suggests that, in the absence of countervailing policy intervention, middle-class Black Americans may not share in the economic benefits of these new technologies; in fact, many Black workers may be at risk of economic harm.

What We Found

We find that Black workers are, by some measures, overrepresented in roles that are relatively less exposed to AI, a pattern driven in part by the association between AI exposure and educational attainment and that may reflect occupational segregation. We also find that Black middle-class workers are somewhat more exposed to AI than Black workers overall. Finally, we find evidence that, among occupations that are more exposed to AI, Black workers tend to be overrepresented in roles that may be at greater risk of automation and displacement.

These early findings suggest directions for both further research and policy responses to ensure that Black workers are protected from the risks and share in the benefits of AI. While further research is necessary to better understand a number of these issues, available evidence suggests the following directions for policy action now:

• Enforce equitable access to employment, offer stable employment, and guarantee fair compensation. For example, investigating instances of worker displacement based on race and gender can help ensure that Black workers are not discriminated against or disadvantaged in the hiring process or while working.
• Address potential economic impacts of AI on Black workers by drawing on research on how policy can most effectively protect workers’ economic security, regardless of the specific economic or technological forces that might lead to their displacement.
• Better support all workers, including Black workers, by expanding access to retraining and reskilling programs. By investing more and proportionate resources in mentorships, apprenticeships, and technical training programs, workers can learn how to use new technologies, including AI, and navigate job changes and the creation of new roles.

How We Did It

To investigate the degree of AI exposure among Black workers and its potential implications, we conduct an analysis that combines data on AI exposure at the occupation level with data on worker demographics and labor force details. We use the relative level of educational attainment across occupations, in conjunction with differences in AI exposure, as a rough proxy for the degree to which AI poses varying risks of displacement.

While the AI exposure measure we employ has been commonly used in broadly similar analyses, it does not reflect current or actual adoption or use of AI at work. In one sense, this is a strength of the measure for considering the likely future impacts of AI, given that AI adoption is still in relatively early stages and that forward-looking questions are of interest. For this same reason, however, it is necessarily prospective and uncertain. Only time will tell whether exposure translates into adoption or economic impact.