An NYU professor who hates that students' work reads like McKinsey memos held AI oral exams to 'fight fire with fire' (www.businessinsider.com)

An NYU professor has implemented AI oral exams in response to student submissions that increasingly resemble polished consulting memos, lacking genuine comprehension of the material. Panos Ipeirotis, teaching data science at NYU's Stern School of Business, turned to AI technology to scale oral examinations—historically deemed too cumbersome for large classes. By utilizing an AI agent developed with ElevenLabs' conversational speech capabilities, Ipeirotis sought to enhance assessments that gauge understanding and reasoning, ultimately presenting a solution to the challenges posed by AI-informed coursework. The oral exam format allowed the AI to interrogate students about their projects and apply real-time reasoning to class discussions over several days, costing a mere $15 for 36 students—far less than traditional human-led assessments. Additionally, Ipeirotis employed three AI models—Claude, Gemini, and ChatGPT—to grade the results, achieving greater consistency and improved feedback than human assessors typically provide. While students had mixed reactions, with many expressing anxiety over the new format, the initiative underscores a critical evolution in educational assessment as institutions strive to balance traditional learning methods with AI technologies that reshape plain evaluations and promote genuine understanding.