A professor in Mississippi recently caught 32 out of 35 students using artificial intelligence to cheat on a midterm exam. Jason Gibson, who teaches at Alcorn State University, shared the results in a viral social media series. To identify the dishonest work, Gibson included a trap in his assignment prompt. He inserted hidden white text that instructed students to include the word Madagascar in their answers, even though it had no relevance to the Industrial Revolution topic he assigned.
Most of the submitted essays contained the word, proving that the students relied on automated generation rather than their own writing. When confronted, many students failed the exam. Only two students attempted to contest their grades, with one claiming she saw the hidden text while using dark mode on her screen, though the academic repercussions stood.
Gibson stated that his intent was not to punish students but to highlight the robotic nature of AI-generated prose. He noted that such tools often strip personality from writing. While some students ended up with failing or poor marks, others earned passing grades, and the instructor ultimately moved on from this specific method of detection.
This incident highlights a growing tension in higher education. As schools grapple with the prevalence of AI, faculty members are split between policing academic integrity and teaching students how to work with these new tools. Experts suggest that rather than focusing solely on catching cheaters, institutions should update assignments to reflect the current technical environment. Gibson himself has since ceased using this specific detection method, as he shifts his focus toward traditional classroom instruction.