A University of Chicago study presented new evidence that administrators may not have reliable information about how widely AI tools are being used by students. In an anonymous survey of 338 undergraduates, 60% reported personal use of AI tools such as ChatGPT, while 90% said they believed the average student uses AI. Researchers tied the 30-point discrepancy to social desirability bias—students may underreport their own usage out of concern about being judged—or to peer estimation effects, where visible AI activity on campus makes AI adoption seem more universal than it is. The study’s authors warn that without better measurement, colleges risk building AI policies on inaccurate assumptions, complicating academic integrity enforcement and course design decisions.