When AI Screens AI, Hiring Bias Hides in Plain Sight
An opinion column by Dr. Gleb Tsipursky warns that AI screening tools can rank equally qualified candidates differently depending on which writing model they used. It cites a September 2026 simulation of eight writing models and six evaluator models across ten jobs, where in the worst pairing about two of every 25 qualified applicants dropped out, and University of Maryland research finding evaluators favored resumes from their own model family. The column urges HR to test matched candidate profiles, review rejected applicants, and ask vendors what drives scores.
