A model predicts AI-assisted applications can raise screening barriers.

What the paper models

The authors model a hiring market in which applicants can use AI to generate and tailor application materials. That can make it easier to apply, but it can also make the materials less informative about a person's fit for a job. Employers must decide whom to screen when screening is costly.

There are no jobseekers, firms, vacancies, interviews or hires observed in this paper. Its result follows from the assumptions in the model.

What the model predicts

If application materials become less informative, an employer can rationally put more weight on a coarse signal: prior experience. In the model, an inexperienced applicant who would be a good fit is the group most likely to lose a screening opportunity. The problem is not that the applicant is necessarily less capable. It is that the employer has less credible evidence to tell applicants apart.

The model also identifies a possible response. A lower-cost intermediate assessment can supply new evidence of fit before a more expensive interview or full screening step.

What it cannot tell us

This is a first-version theoretical preprint. It does not estimate how often people use AI in applications, whether employers can detect it, whether experience is receiving more weight, or whether any assessment improves hiring. It does not measure wages, employment, layoffs or unemployment.

A plausible mechanism is not an observed labor-market effect. It should not be read as proof that AI has made hiring worse for new entrants.

How it compares with other evidence

Recent U.S. studies report narrower, exposure-based signals: fewer junior postings in some highly AI-exposed work, and lower early employment or earnings in some graduate data. They do not identify this application-screening channel. Yale's broad U.S. monitor still finds no clear economy-wide AI labor-market footprint.

The new paper gives researchers a sharper question to test: when application assistance rises, do interview and offer rates change differently for inexperienced applicants after employers account for actual skills and prior experience?

A fast claim check

When a headline says AI is making entry-level hiring harder, ask what it measured. A theory can identify a channel. A survey can describe what people report. A hiring record can show who received an offer or job. Those are different claims.