Brown professor's proctored exam reveals AI-dependent students lack foundational skills
A Brown University economics professor's shift from take-home to proctored exams exposed a stark performance gap, with class averages plummeting from 96 to 48.6 percent. Corroborating research from UC Berkeley and Chinese institutions demonstrates that students relying on AI for homework struggle significantly when assessed under supervision. This pattern signals a critical vulnerability in AI-augmented learning: without structural constraints, students may outsource cognitive work rather than develop competency, creating a hidden competency debt that surfaces only during high-stakes evaluation. The finding raises urgent questions about how institutions should calibrate AI access in coursework to preserve learning outcomes.
Modelwire context
ExplainerThe more precise framing here isn't about cheating. It's about a structural mismatch between how students are being assessed and how they are actually learning. A 96 percent average on AI-assisted work isn't inflated grades so much as it's a measurement of a different skill entirely, one that disappears the moment the tool does.
This story is largely disconnected from recent activity in our archive, which has focused primarily on model capabilities and deployment. It belongs instead to a quieter but growing body of work on AI's second-order effects on human skill formation. The relevant space here is educational psychology and institutional policy, not product releases. What makes this finding worth tracking is that it provides concrete, quantified evidence for a concern that has mostly circulated as anecdote. The UC Berkeley and Chinese institutional data cited alongside the Brown case suggest this isn't one professor's outlier result but a reproducible pattern across different academic contexts.
Watch whether major universities announce changes to exam policy or AI-use disclosure requirements before the 2026 fall semester. If institutions respond with structural constraints rather than honor-code updates, that signals the competency-debt framing is gaining traction at the administrative level.
This analysis is generated by Modelwire’s editorial layer from our archive and the summary above. It is not a substitute for the original reporting. How we write it.
MentionsBrown University · UC Berkeley · China
Modelwire Editorial
This synthesis and analysis was prepared by the Modelwire editorial team. We use advanced language models to read, ground, and connect the day’s most significant AI developments, providing original strategic context that helps practitioners and leaders stay ahead of the frontier.
Modelwire summarizes, we don’t republish. The Decoder originally reported this story as “Grades dropped from 96 to 48 percent when a Brown professor made students take the exam without AI”. The full content lives on the-decoder.com. If you’re a publisher and want a different summarization policy for your work, see our takedown page.