Grades dropped from 96 to 48 percent when a Brown professor made students take the exam without AI

Editor's take

A Brown University economics professor observed a dramatic drop in student exam scores, from an average of 96% on a take-home assignment to 48% on a subsequent in-person final, strongly suggesting widespread AI-assisted academic dishonesty. This incident highlights the growing challenge educational institutions face in maintaining academic integrity as generative AI tools like ChatGPT become more accessible and powerful, potentially devaluing traditional assessments and requiring a significant reevaluation of pedagogical approaches.

The implications extend beyond Brown, impacting universities globally and prompting discussions about how to adapt to AI's presence in learning environments. Future developments will likely focus on the effectiveness of new proctoring technologies, the integration of AI detection software, and the development of assessment methods that are inherently resistant to AI manipulation, such as oral examinations or project-based learning that emphasizes original thought and critical application.