A warning about student AI use lowered hiring ratings, not actual hires

What the study measured

Boyd-Swan and Reynolds recruited 1,750 U.S. full-time, for-profit-sector professionals with hiring responsibilities through Prolific in May and June 2026. They randomly showed either no added AI message, a milder message about AI helping students, or a message stressing that AI can replace student effort and ability. Each participant rated two fictional recent-graduate resumes, one with a high GPA and one with a low GPA.

Participants scored how likely they would be to interview each applicant on a seven-point scale and what percentage above an initial salary offer they would be willing to negotiate. These are stated intentions in a survey, not observed interviews or pay.

What changed

The stronger warning reduced ratings for interviews and salary negotiation, especially for high-GPA resumes. For white-male resumes, the high-GPA interview-rating advantage over low-GPA resumes fell from 1.04 points with no added message to 0.62 points with the strong AI message. The mild message did not produce a clear general premium. The authors warn that some subgroup differences are imprecise.

Random assignment supports a causal conclusion about how the message changed this sample's survey ratings. It does not prove that AI use in college reduced ability, that employers currently hold this belief at the tested level, or that hiring has fallen.

What it cannot answer

The respondents were a selected online sample evaluating generic resumes, not an employer-representative sample filling real vacancies. The study did not observe interview invitations, offers, hires or realized wages. The salary figure is willingness to negotiate, not a pay change. A prior London field experiment measured real interview invitations from an AI-related credential signal; it did not test this belief frame. The two results do not cancel or confirm each other.