New York University’s vice provost advocates a return to oral examinations, in-class essays, and real-time assessments to address widespread generative AI cheating. Traditional plagiarism detection tools fail to reliably identify AI-generated work, prompting educators to reconsider assessment formats. This shift echoes medieval pedagogical practices emphasizing oral instruction over written assignments. Although introducing challenges for large classes and some students, these measures aim to require active student engagement and authentic knowledge demonstration, signaling a major pedagogical adaptation amid evolving AI impacts on academic integrity.