AI Hiring Tools Yield Racial Bias and Systemic Rejection; 26% Black & 15% Asian (hai.stanford.edu)

A large-scale study has revealed troubling racial bias in AI hiring algorithms, which could have major implications for job seekers in a challenging labor market. Analyzing 4 million job applications from 3.4 million people, researchers found that 26% of Black applicants and 15% of Asian applicants faced discrimination due to AI screening tools, which are widely used by 90% of U.S. employers. The study highlights that traditional metrics for measuring discrimination may obscure systemic bias, as averaging recommendations across job types can mask the adverse impact on specific racial groups. For instance, if Black and Asian candidates were recommended as often as white candidates, 40,000 additional applications for these groups would have progressed in the hiring process. Moreover, the research indicates that reliance on a single vendor for hiring algorithms creates an “algorithmic monoculture,” increasing the chances of systemic rejection for applicants submitting multiple applications. This means individuals using the same AI tool across various job applications are more likely to be rejected from all roles, exacerbating the likelihood of limited job access based on race. The findings emphasize the urgent need for independent research and regulatory oversight in AI hiring practices to ensure fairness and transparency in the increasingly automated employment landscape.