Signal Can Labor Markets Function in the Age of AI? The Evaluation Bottleneck in Hiring

Summary

The paper studies how AI-assisted job-search tools, which make it easier to generate and tailor applications, can reduce how informative application materials are about applicant fit. In a hiring model where applicants differ in experience and latent match quality, firms use noisy materials to decide whom to screen. As materials become less informative, a Bayesian firm relies more on coarse observables such as prior experience. Inexperienced-compatible applicants are the most exposed because they lack observable experience and lose individualized information. When screening is costly, firms may screen no applicants or only experienced ones. Multistage hiring with a cheap intermediate assessment can arise endogenously and restore screening opportunities for high-fit workers without prior experience.