Calibration labour: making careers machine-legible under algorithmic hiring – Information, Communication and Society

‘This article reconceptualises algorithmic hiring as a transformation in the communicative conditions of recruitment rather than primarily as a problem of decision-making. Drawing on two years of ethnographic fieldwork at a London-based employment agency and training programme, it examines how employability is produced when recruitment is mediated by AI-enhanced applicant tracking systems (ATS). The article introduces algorithmic legibility as a communicative regime through which candidates encounter evaluation via asymmetrical proxy signals – such as metrics, thresholds, and prolonged silence – rather than through dialogic explanation. It demonstrates how this shift transforms job seeking from an interpersonal imitation game rooted in socialised modelling into a rationalised calibration game that demands calibration labour: the iterative, individualised restructuring of CVs, timelines, and narratives to align with opaque infrastructural criteria. The article argues that this infrastructure reconfigures recruitment into a restrictive recognition game in which visibility is increasingly reserved for biographies that can be readily rendered machine-legible. Under these conditions, interpretive labour is redistributed downwards, institutions are relieved of explanatory accountability, and marginalised jobseekers are compelled to become technicians of their own datafied subjectivities. Beyond corporate selection, this shift erodes the status of employment as a locus of social recognition and civic participation, contributing to the production of the algorithmic citizen – a subject whose socio-political inclusion becomes increasingly contingent upon continuous machine legibility. Algorithmic hiring, the article argues, governs access to work not through selection alone, but by reshaping whose lives can be rendered recognisable, on what terms, and through what forms of labour.’

Link: https://www.tandfonline.com/doi/full/10.1080/1369118X.2026.2715575