Not yet human-reviewed

This source record was written automatically by the research pipeline and has not yet been reviewed by the maintainer or integrated into a topic page. Treat it as a provisional summary and verify claims against the original source.

Algorithmic Monocultures in Hiring

Citation. Bommasani, R., Bana, S. H., Creel, K. A., Jurafsky, D., & Liang, P. (2026). Algorithmic Monocultures in Hiring. Proceedings of the 2026 ACM Conference on Fairness, Accountability, and Transparency, 6351–6382. https://doi.org/10.1145/3805689.3812400

Summary. This FAccT paper examines whether employers’ reliance on algorithms from the same vendor concentrates adverse hiring outcomes across applicants and racial groups. Using applications screened by one vendor’s algorithms, the authors find racial disparities and outcome homogeneity that exceed chance expectations, and use deterministic replication to model how broadly applicants would need to apply to reach human consideration.

Key claims

• In a dataset of 3 million applicants and 4 million applications screened by algorithms from one vendor, 14.74% of Asian applicants’ applications and 25.87% of Black applicants’ applications were to positions that adversely affected those groups under U.S. employment-discrimination standards.
• Among applicants who applied to 10 positions, 4% were recommended for rejection by every position, a rate the authors report is higher than expected by chance.
• Deterministic replication of the hiring algorithms’ outcomes suggests that applicants must apply widely to increase the chance that a human considers their applications.

Evidence & limitations

• The study analyzes a novel large-scale observational dataset and uses deterministic replicability to generate counterfactual outcomes across positions; the available Jina artifact contains the abstract and references rather than the paper’s full methods and results text.
• The source is a peer-reviewed FAccT 2026 proceedings paper; its inferences concern the studied vendor’s algorithmic screening context and should not be generalized to all hiring systems without further evidence.

Feeds

• algorithmic-assessment
• inclusive-hr-systems
• Proposed new topic: algorithmic-monocultures-in-hiring — shared hiring vendors can concentrate correlated individual and racial adverse outcomes across employers.