- June 23, 2026
- By Gregory Muraski
As artificial intelligence plays an increasingly major role in hiring, new research from the University of Maryland’s Robert H. Smith School of Business shows that when evaluating candidates, AI systems rate resumes they generated more favorably than those written by humans—or even by competing AI models.
In a new working paper, Smith Ph.D. candidate Jiannan Xu, along with Gujie Li Ph.D. ’25 of the National University of Singapore and Jane Yi Jiang Ph.D. ’24 of Ohio State University document how large language models used in hiring may systematically prefer resumes generated by the same model.
Their experiment demonstrating this “self-preference bias” was based on more than 2,200 resumes and leading AI models. The findings have drawn coverage from outlets such as the New York Post and Business Insider.
Across major commercial and open-source models, the study reports self-preference rates ranging from roughly 67% to 82% when comparing AI-generated resumes with human-written ones. The researchers interpret this pattern as evidence that model-specific writing styles may influence evaluation outcomes, independent of applicant qualifications.
The researchers simulated hiring pipelines across 24 occupations, and found that candidates using the same AI system as an employer’s screening tool have a 23% to 60% higher likelihood of being shortlisted versus equally qualified candidates submitting human-written materials.
Disparities were especially pronounced in business-related roles such as sales and accounting, where standardized language and formatting may amplify similarities between AI-generated resumes and evaluation criteria.
Xu, who is affiliated with UMD’s Institute for Trustworthy AI in Law & Society and the Maryland Language Science Center, said the study highlights an entirely new form of bias.
“Hiring is an early example, but these interactions are likely to become much more common as AI tools increasingly create, screen, rank, and evaluate information across society,” Xu said. “That makes AI-to-AI bias an important new frontier for fairness research and governance.”