NextFin News - The assignment involves no laptop, no chatbot, and no technology of any kind. In fact, there is no pen or paper either. Instead, students in Chris Schaffer’s biomedical engineering class at Cornell University are now required to speak directly to an instructor in what he calls an "oral defense." This testing method, as old as Socrates, is making a sudden comeback across American higher education as professors grapple with a crisis of authenticity triggered by generative artificial intelligence. Educators are reporting a troubling new phenomenon: take-home essays and problem sets are coming back perfect, yet when those same students are asked to explain their work in person, they are met with blank stares.
The shift toward oral assessments marks a fundamental pivot in how universities measure human intelligence in an era where machines can mimic it. At the University of Pennsylvania, Emily Hammer, an associate professor of Middle Eastern Languages and Cultures, has begun pairing oral exams with written papers. Hammer notes that the move isn't merely about policing academic integrity; it is a response to a visible decline in cognitive capacity and creativity among students who increasingly view the hard work of thinking as an optional step that can be outsourced to a prompt. According to Bruce Lenthall, executive director of Penn’s Center for Teaching and Learning, the Ivy League institution is seeing a "massive shift" toward in-person assessments, with faculty workshops now dedicated to mastering the art of the viva voce.
While oral exams have long been a staple of the Oxbridge tutorial system in England, they have historically been rare in the modern American undergraduate experience. The logistical hurdle of testing hundreds of students individually has traditionally favored the efficiency of written Scantrons and essays. However, the launch of ChatGPT in late 2022 has rendered those efficiencies a liability. At New York University, Clay Shirky, vice provost for AI and technology in education, reports that faculty are increasingly resorting to cold-calling and mandatory office hours to verify student knowledge. The consensus among administrators is becoming clear: if a student cannot defend a thesis face-to-face, the university can no longer assume the student actually understands the material.
Some institutions are attempting to solve the scalability problem by using the very technology they are fighting. Panos Ipeirotis, a professor at NYU’s Stern School of Business, recently deployed an AI-powered oral exam. Students interact with a voice-cloned agent that drills them on their projects, providing clues or criticism based on their responses. While student feedback has been mixed—some find the experience of talking to a blank screen "jarring"—Ipeirotis argues that the traditional written assignment can no longer be trusted as a proxy for actual thinking. This "fighting fire with fire" approach suggests a future where AI acts as both the tutor and the proctor, leaving humans to focus on the high-stakes synthesis of information.
The stakes for this pedagogical retreat are particularly high in STEM and professional disciplines. In fields like biomedical engineering or computer science, skipping the mental struggle of problem-solving creates a hollow foundation that collapses in upper-level courses or professional practice. Schaffer’s 20-minute Socratic sessions at Cornell are designed to "incentivize" students to do the work themselves. By shifting the grading weight from the written output to the oral defense, the university is effectively devaluing the "product" of education—the paper—and revaluing the "process"—the student's ability to articulate technical knowledge under pressure.
Critics of the oral revival point to the potential for bias and the anxiety it induces in shy or neurodivergent students. Carolyn Aslan, who leads oral exam training at Cornell, acknowledges these hurdles but suggests that one-on-one breakthroughs often happen with the quietest students when they are forced out of the anonymity of a large lecture hall. For students like Olivia Piserchia, a junior at Cornell, the "live check-in" provides a level of accountability that written work lacks. It is significantly harder to look an instructor in the eye and admit ignorance than it is to submit a chatbot-generated paragraph. As universities recalibrate, the oral exam is emerging not as a relic of the past, but as a necessary filter for the future of human expertise.
Explore more exclusive insights at nextfin.ai.
