Brown University Professor Suspects AI Cheating in Exams, Average 96% in Take-Home Midterms, 48.6% in In-Person Finals
A professor at Brown University suspects that students used AI to cheat on a take-home midterm exam, which had an average score of 96%. This was followed by an in-person final exam, where the average score dropped to 48.6%.
Source: Public Information
ABAB AI Insight
This incident reflects the adjustments in examination formats at universities due to the widespread use of AI tools. Similar cases have previously been seen in the challenges of academic integrity in remote assessments.
From a capital perspective, AI-assisted learning boosts take-home exam scores, while in-person assessments return to evaluating real abilities, motivated by the need to maintain academic integrity and strategically push educational assessments from memory-based to application-based capabilities.
Many universities are deploying AI detection tools and reforming exams, as the U.S. is currently in the early stages of transitioning higher education from traditional assessments to a balance of integrity and capability in the AI era.
Essentially, this is a case of technological substitution: AI cheating distorts score distributions, stemming from the ease of use of generative tools, prompting a shift in pricing power from open take-home assessments to supervised in-person evaluations and reshaping the structure of educational equity.
ABAB News · Cognitive Law
96% at home, 48.6% in person; AI cheating exposes real ability gaps.
Assessment formats determine the boundaries of integrity; in-person exams reshape academic evaluation standards.
Technological substitution accelerates educational reform, with teachers using data to drive upgrades in assessment systems.