Can Unbiased Predictive AI Amplify Bias?

Tanvir Ahmed Khan ()

No 1510, Working Paper from Economics Department, Queen's University

Abstract: I analyze a model of unbiased predictions mediated by biased humans, showing that a bias-neutral precision gain (e.g. from AI adoption) is not generally bias-neutral in its effects. Expected victims of bias are discriminated-group applicants with qualifications warranting approval under a counterfactual unbiased standard but not the stricter biased standard they face. For them, precision raises discrimination by reducing noise-driven chance approvals. Overall discrimination can also increase when precision shifts more realized predictions of discriminated-group applicants into than out of the between-standards interval. This isolates a new channel through which AI can exacerbate discrimination, distinct from biased design or data.

Keywords: AI-assisted decision-making; Human bias; Prediction Precision; Human-Machine Interaction; Discrimination; Statistical Discrimination; Algorithmic Bias; Decision-Making under Uncertainty; Human Discretion (search for similar items in EconPapers)
JEL-codes: D63 D83 J71 O33 (search for similar items in EconPapers)
Pages: 23 pages
Date: 2026-07
New Economics Papers: this item is included in nep-ain, nep-big and nep-cmp
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
https://www.econ.queensu.ca/sites/econ.queensu.ca/files/wpaper/qed_wp_1510.pdf First version 2026 (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:qed:wpaper:1510

Access Statistics for this paper

More papers in Working Paper from Economics Department, Queen's University Contact information at EDIRC.
Bibliographic data for series maintained by Mark Babcock ().