One in Four Job Applications from Black Candidates Hit by Adverse-Impact Screening Algorithm

A Stanford and Northeastern study of 3 million applicants and 4 million applications screened by a single vendor's algorithm found that 25.87% of Black applicants' submissions landed in roles where the system produced adverse impact under U.S. employment discrimination standards. For Asian applicants, the figure was 14.74%. The researchers reconstructed outcomes across positions applicants never applied to, revealing correlated rejections invisible in traditional audits.