Recruiters in the Loop: Hybrid Agency and Sociomaterial Tensions in AI-Mediated Hiring Practices

The increasing integration of artificial intelligence within organisational life has renewed debates surrounding the transformation of professional work and, more fundamentally, of professional identity. Rather than treating such transformation as technologically determined, this paper draws on a sociomaterial perspective and STS scholarship to examine how AI-driven tools are appropriated within recruitment work — a domain in which algorithmic rationalities and professional judgment are in persistent, often uneasy, tension.
The empirical basis of this study consists of seventeen semi-structured interviews with recruiters who actively use AI recruitment tools across varied organisational contexts. The decision to utilise interviews as a distinct analytical entry point was made with the understanding that they would not be a substitute for interactional data. Retrospective accounts have been found to reveal the sense-making work through which practitioners reconstruct their engagement with AI, legitimate their professional choices, and negotiate their agency. This methodological positioning engages a broader epistemological problem — how human–AI interaction can be studied when direct observational access is foreclosed by proprietary systems and organisational opacity.
Three sociomaterial tensions emerged inductively from the analysis: the first one emerges between the objectives of workload reduction and its intensification; the second one emerges between process and information standardisation and the demands for personalisation; and the last one emerges between the use of AI as a support to professional expertise and the potential of AI to substitute to that expertise. Questions of accountability were found to be of particular significance in all three tensions, with the question being posed of who should be held responsible for making consequential decisions when agency is distributed across human and computational actors. Rather than merely adopting or rejecting available tools, recruiters develop hybrid adaptation strategies through which technological logics — efficiency, automation, standardisation — are selectively articulated with professional logics of judgement, relationality, and contextual sensitivity, progressively reconfiguring the rationalities embedded in AI systems.
The paper makes a theoretical contribution by moving beyond the binary of acceptance and resistance towards a more fine-grained account of professional agency under algorithmic constraint. It also makes an empirical contribution by proposing a typology of sociomaterial tensions specific to hiring work. Furthermore, it reflects on the epistemological challenges of studying human–AI interaction through retrospective accounts.