AI Resume Screening Audits Reveal 'Fairness' Can Mask Incompetence, Study Finds
A new arXiv preprint audits eight AI-powered resume screening platforms and finds that some systems appear unbiased only because they fail to meaningfully evaluate candidate qualifications. The study demonstrates that models can show apparent demographic fairness while lacking the competence to distinguish relevant from irrelevant experience. The authors propose a dual-validation framework, arguing that both bias and evaluative competence must be audited before deploying AI in hiring.
Why it matters: The findings highlight a critical oversight in current AI hiring audits, showing that focusing solely on fairness can allow unqualified systems to be deployed, with direct implications for responsible AI use in employment decisions.
Full story at: arXiv Computers and Society ↗