Detecting AI-Generated Essays in Writing Assessment: Responsible Use and Generalizability Across LLMs
The episode explores Jiangang Hao's research on detecting AI-generated essays in standardized testing. The hosts discuss the limitations of current detectors against evolving models like GPT-5 and the benefits of a 'GPT-all' training approach. They conclude that effective detection requires combining text analysis with writing process data, such as keystrokes.