Year: 2026
Type: conference-paper
Abstract: Generative artificial intelligence (AI) is widely expected to reshape human resource management (HRM), yet organization-level evidence on how it redistributes work between people and machines is scarce, and debate is split between optimistic augmentation and pessimistic attrition narratives. This paper asks a narrower, answerable question: after an enterprise generative AI platform is deployed in HR, which tasks move to the machine, which remain human, and how do practitioners experience the change? We report an exploratory pre–post study of a single platform deployed across ten organizations over six months, combining anonymized operational data (more than 6,000 requisitions) with thirty-two practitioner interviews. The design is observational, with no control group, so results are reported as associations rather than causal effects. Time-to-hire fell by a median of 23% across sites (95% CI 20-26%; from about 42 to 32 days), and the volume of routine queries reaching human advisers fell by 72% (95% CI 69-75%); every site moved in the same direction (Wilcoxon signed-rank p = 0.002, matched-pairs rank-biserial r = 1.00), though with only ten sites this is indicative. Both varied across sites and accrued gradually over the six months rather than at switch-on. The platform absorbed high-volume, rule-bound tasks such as screening, scheduling, and policy queries, while judgment- and relationship-based tasks remained human. Practitioners reported reallocating time towards that relational work, alongside concerns about transparency, surveillance, and new supervisory demands. We do not observe long-run employment outcomes: no participating organization reduced HR headcount during the window. Six months is shorter than a typical restructuring cycle, and we cannot establish that augmentation, rather than attrition, is the lasting result. We therefore treat the optimistic reading as a conditional leadership choice, not a finding. Because the platform is a commercial product connected to the authors and the data are vendor-context, we disclose this relationship in full and report the study as early, transparency-qualified evidence. The contribution is empirical: a rare multi-site, mixed-methods account of task redistribution after a generative AI HR deployment, offered with explicit attention to its limits and to the governance obligations, including high-risk AI rules, that such deployments now entail.
Author: Rohini Krishna Kaninde
Institution: Clark University
Language: English
Topic: AI and HR Technologies
Domain: Social Sciences
Open Access status: diamond
OpenAlex ID: W7223946118