Grades dropped from 96 to 48 percent when a Brown professor made students take the exam without AI
A Brown University economics professor removed AI tools from a take-home exam, causing the average grade to drop from 96% to 48.6%. Nearly a quarter of students withdrew from the course.
- An economics professor at Brown University found that removing AI tools from an exam caused average grades to drop from 96% to 48.6%.
- Nearly 25% of students withdrew from the course or skipped the final exam when AI was banned.
- Studies from China and UC Berkeley support the claim that AI dependency reduces performance on proctored exams.
- The experiment raises questions about academic integrity and the future of unsupervised assessments in education.
An economics professor at Brown University suspected widespread AI-assisted cheating during a take-home exam that averaged 96% across 86 students. To test his hypothesis, he administered the final exam in person without AI access. The results were striking: the average score fell to 48.6%, and 18 students dropped the course while 9 did not attend at all.
The professor's informal experiment aligns with broader research from China and UC Berkeley, which suggests that students who rely on AI tools for homework and assignments perform significantly worse on proctored exams. These studies indicate a growing dependency on AI among students, raising concerns about academic integrity and the effectiveness of unsupervised assessments.
The incident highlights the challenges universities face in adapting to AI tools in education, particularly as students increasingly turn to generative AI for coursework. It also underscores the need for institutions to rethink exam design and assessment methods in an era where AI assistance is readily available.
Students relying on AI for coursework may struggle in proctored settings, impacting academic performance.
Raises concerns about AI's role in education and the reliability of unsupervised assessments.
BusinessClaude Cowork's biggest use case is the mundane office work nobody wants to own, Anthropic says
Anthropic analyzed 1.2 million Claude Cowork sessions and found that half of all usage is dedicated to administrative tasks like reports and slide decks, while developers prefer Claude Code.
BusinessOpenAI CEO Altman is now "pretty sure" AI is net job-creating, which is quite the pivot from predicting mass layoffs
OpenAI CEO Sam Altman now believes AI is net job-creating, reversing earlier warnings of mass layoffs. Anthropic's CEO has also softened similar predictions.
Meet SK Hynix, the Key Nvidia AI Partner That's Delivering Triple-Digit Growth and Just Launched on the Nasdaq. Is the Stock a Buy? - The Motley Fool
SK Hynix, a critical Nvidia AI chip supplier, has listed on the Nasdaq, reporting triple-digit revenue growth driven by AI demand.
Getting campaign text messages ahead of midterms? There could be an AI bot behind it - WXXI News
AI chatbots are being used to send campaign text messages to voters ahead of the midterm elections, raising transparency and regulation questions.
Trump administration targets state AI laws over ideology - Jefferson City News Tribune
The Trump administration is moving to override state AI laws, citing ideological concerns rather than technical or safety reasons.
India's Tata Consultancy Services plans up to 8,900 AI deployment engineers, seeks AI acquisitions - Reuters
Tata Consultancy Services plans to hire up to 8,900 AI deployment engineers and is seeking AI acquisitions. This move is part of the company's efforts to expand its AI capabilities.