world
Study Finds AI Applicant Systems Favor Resumes Written by Same AI Model
A study of AI-powered applicant tracking systems found they selected resumes generated by their own large language model 23% to 60% more often than human-written resumes. Researchers tested multiple models across 24 occupations and identified the strongest effects in accounting, sales and finance.
May 16, 8:11 AM·1m read1 source
nypost.comAudio version
Tap play to generate a narrated version.
Org in February. 1. Researchers then simulated hiring pipelines for 24 occupations and measured selection rates.
AI evaluators were 23% to 60% more likely to advance candidates whose resumes matched the employer’s model. The effect appeared most pronounced in accounting, sales and finance. Researchers Jiannan Xu of the University of Maryland, Gujie Li of the National University of Singapore and Jane Jiang of Ohio State University conducted the work.
They stated that the pattern constitutes a novel form of bias that can advantage candidates using the same LLM while disadvantaging equally qualified applicants who do not.
Professor Emma Wiles, who studies information systems and AI’s impact on the labor market, said the systems may select applicants that match the AI’s own style rather than those who best demonstrate ability. Wiles recommended that job seekers use AI tools to refine their own writing rather than replace it.
She stated that such assistance can improve clarity without removing the applicant’s individual voice. More than 300,000 job cuts were announced between January and April 2026, with notable reductions in the tech sector, according to research firm Challenger, Gray & Christmas.
