Kolloquium: Hiring Under Technological Change: AI Disclosure and Pre-Screening Sorting in the Applicant Pool

Abstract:
We study how employer AI disclosure in actual job vacancy posts alters applicant behavior and labor supply prior to employer screening. In addition we document the important role of employer offered AI training in mitigating the substantial adverse effects of AI mention on applications. To do this we leverage a randomized controlled field experiment tracking 3,865 live applications across 1,456 posting-days at a global industrial firm (Siemens), complemented by a linked survey experiment (N = 5,192). Our main findings are: Disclosing AI use reduces application volume by roughly one-third. Application drops are heavily concentrated in Coordination roles, moderate in Technical Specialist positions, and negligible in Direct Production. Standard AI exposure indices do not predict this pattern; deterrence instead scales with the overlap between AI capabilities and core interpersonal tasks. Candidates without AI-relevant skills are disproportionately deterred, shifting the composition of the applicant pool before employer evaluation begins. Conditional progression to shortlisting and hiring remains stable. In terms of mechanisms we document that AI disclosure acts as applicant-facing signals of job reorganization, task uncertainty, and costly adaptation rather than mere productivity enhancements. However, coupling AI disclosures with explicit commitments to employer-provided AI training substantially weakens deterrence, mitigating perceived role ambiguity and displacement concerns.