AI-Written Exam Answers Outperform Human Submissions in University Study

University of Reading research reveals AI-generated exam answers often go undetected and receive higher grades than those by actual students.

94% of AI-generated exam submissions were not identified as such by human markers.
AI-written answers scored higher than real student submissions in over 83% of cases.
Study highlights challenges in maintaining academic integrity with the rise of generative AI.
Researchers suggest educational institutions must adapt to the evolving presence of AI.
Calls for revising assessment methods to address AI's impact on academic evaluations.