A framework for ethical AI-HRM development and use

Abstract Our work addresses the challenge of integrating artificial intelligence (AI) into human resource management (HRM) systems while minimizing bias, increasing transparency, and ensuring accountability. We examine how developers and adopters of AI-HRM systems can proactively identify and mitigate bias while adhering to U.S. anti-discrimination laws. In this paper, we propose a framework for the evaluation of equity and transparency in AI-HRM systems developed through an analysis of U.S. laws related to discrimination and a parallel examination of established ethical frameworks for AI. Our work reveals shortcomings in the transparency and equity of existing AI-HRM systems, highlighting the pervasive nature of potential biases and the ambiguity surrounding accountability. These findings underscore the need for a structured approach to bias evaluation, directly supporting the necessity and value of the proposed bias audit benchmark as a tool for ensuring responsible AI adoption in HRM. While prior works highlight technical obstacles in developing bias-free AI-powered systems, they fail to provide technical solutions that incorporate legal and ethical ramifications. Our work bridges the gap by explicitly addressing issues of AI bias within HRM systems from technical, legal, and ethical perspectives. In particular, we synthesize the “siloed” approach of these different areas of work into a unified path forward grounded in existing theoretical principles.