Abstract

AI is currently extensively deployed in recruitment technologies with little regulation or guidelines either from within corporations or from external regulatory bodies. We’re seeing a huge demand among HR practitioners for AI-powered software that reduces HR costs and streamlines the recruitment process, while simultaneously allowing them to achieve corporate DEI goals. However, recruitment stakeholders often have a limited understanding of how AI-powered tools work. In this paper, we look at how gender studies and critical race theory can interrogate some of the claims made by Recruitment AI companies about how their tools help mitigate bias, and demonstrate why practitioners should be attentive to how difference is produced rather than making futile attempts to remove it altogether. We discuss four limitations of anti-bias measures in AI for hiring: 1) debiasing measures treat race and gender as isolatable characteristics or attributes, rather than systems of power; 2) outsourcing ‘diversity work’ to technologies may ignore or distract away from systemic inequalities in organisations 3) AI-powered hiring tools replicate the unequal power relations between the observer and the observed; 4) AI hiring software produces the ideal candidate it supposedly identifies. We close with three recommendations regarding the use of AI-powered hiring tools and software.