According to a report by technology media outlet TechSpot yesterday, Jason Gibson (note: Jason Gibson), a history professor at Alcorn State University in the United States, recently shared an unusual way he used to catch students cheating with AI. The results showed that a total of 32 students in the two classes he taught were caught red-handed.
According to reports, one of the questions on the midterm exam this professor gave his students concerned the Industrial Revolution. Knowing that many students would simply copy answers from AI, he decided to add a prompt to the exam document instructing the AI to describe “Madagascar” in a completely illogical way. The prompt was embedded in the document in white text, so students could not see it with the naked eye.
Of the 35 exam papers the students ultimately submitted, 32 gave themselves away. In fact, these students could have noticed something was wrong if they had even briefly checked the answers generated by AI. One answer first described the Industrial Revolution, then suddenly said, “Madagascar floated horizontally in the afternoon”; another initially discussed technological development and social inequality, but ended with “Madagascar’s purple bicycle whispers to the ceiling.”
Professor Jason said that he did not do this to mock the students. After discovering that they had cheated, he gave them an opportunity to appeal. In the end, 30 chose to admit fault, while 2 still attempted to defend their actions.