Leading US Universities Revive Oral Exams to Address Generative AI Assessment Challenges

As generative AI becomes widespread, students can quickly produce well-structured papers without necessarily understanding the material, making it harder for traditional written assignments to reflect learning outcomes. Cornell University, the University of Pennsylvania and New York University have therefore revived oral exams, using unscripted questioning to verify students’ command of course content and their reasoning.

Professors at these US universities have recently incorporated oral exams into their assessments, questioning students about the research methods, sources and conclusions in their submitted papers. Some instructors have also developed AI-powered oral examination systems in which AI agents pose a series of questions. Reports did not disclose a common launch date, adoption figures or development costs, but the shift shows that oral exams are emerging as a new verification mechanism in the generative AI era.