Algorithmic Monocultures in Hiring

2026/05/26 by Rishi Bommasani, Sarah H. Bana, Kathleen A. Creel +2 · 11 voices · 1 citation
#cs.CY #cs.AI

paper · pdf

Abstract

Many employers screen job applicants with algorithms built by the same few algorithm vendors. We hypothesize that algorithmic monoculture leads to the same individuals and members of the same racial groups facing rejection. We acquire and analyze a novel dataset of 3 million applicants submitting 4 million applications where all the applications are screened by algorithms built by the same vendor. We find clear racial disparities in applicant outcomes. Of all applications submitted by Asian and Black applicants, 14.74% and 25.87% are submitted to positions that adversely impact Asian and Black applicants, respectively, according to U.S. employment discrimination standards. Individuals also receive homogeneous outcomes: 4% of all applicants who apply to 10 positions are recommended for rejection from all positions, a rate higher than expected by chance. To better understand this homogeneity, we leverage the deterministic replicability of hiring algorithms to generate the outcomes applicants would have received if they applied to all positions. We show that applicants would need to apply widely in order to ensure their applications are considered by a human

Citations

• AI generates covertly racist decisions about people based on their dialect
• Algorithmic bias: on the implicit biases of social technology
• Discrimination in the Credential Society: An Audit Study of Race and College Selectivity in the Labor Market

Cited by

• The Hitchhiker's Guide to Monoculture

Discussions

• arxiv.org/pdf/2605.27371 [bsky, 41 points, 0 comments]
• Source: arxiv.org/pdf/2605.273... which is the actual paper [bsky, 30 points, 1 comments]
• Link to the paper: arxiv.org/pdf/2605.273... LLM results are dice rolls. So if you rolled a 1, this will retain that result for any hit for that resume (cited here, for 330 days). Changing the resume [bsky, 4 points, 1 comments]
• arxiv.org/pdf/2605.27371 It appears in quoting the news video I saw they got that part slightly wrong (or I misunderstood). Your score for the “pymetrics assessment games” are what they cache. the res [bsky, 3 points, 0 comments]
• Whoa. arxiv.org/abs/2605.27371 So a study out of Stanford found that: - Over 90% of employers are using algorithms for hiring purposes - Most of these algorithms are coming form the same few vendors - [bsky, 2 points, 1 comments]
• 7/ May 2026 - "We acquire and analyze a novel dataset of 3 million applicants submitting 4 million applications where all the applications are screened by algorithms built by the same vendor. We find [bsky, 2 points, 1 comments]
• I think this might be the Stanford paper cited in the two reposts below on LLM AI job application screening: “Algorithmic Monocultures in Hiring” Bommasani, Bana, Creel, Jurafsky, & Liang (2026) arxiv [bsky, 1 points, 0 comments]
• The paper referenced: arxiv.org/abs/2605.27371 An interview with some of the co-authors: digitaleconomy.stanford.edu/news/qa-algo... [bsky, 1 points, 0 comments]
• Looking at the Standford report is pretty interesting. "We analyze data from the hiring algorithm vendor pymetrics from December 2018 through December 2022." arxiv.org/abs/2605.27371 [bsky, 1 points, 1 comments]
• arxiv.org/abs/2605.27371 [bsky, 0 points, 0 comments]
• https://arxiv.org/abs/2605.27371 [bsky, 0 points, 0 comments]

Related

• Algorithmic Hiring and Diversity: Reducing Human-Algorithm Similarity for Better Outcomes
• Fairness and Bias in Algorithmic Hiring: A Multidisciplinary Survey
• Fairness and Bias in Algorithmic Hiring: a Multidisciplinary Survey
• Monoculture in Matching Markets
• Average Profits of Prejudiced Algorithms