History professor uses cunning trick to catch 31 out of 35 students cheating with artificial intelligence

The use of artificial intelligence tools in education remains a double-edged sword. This is clearly demonstrated by a recent case from Alcorn State University in the US state of Mississippi. History professor Jason Gibson attracted a lot of attention online when he revealed how he used a simple test to test the independence of his students in taking a mid-term test.

The results of the test were quite surprising. Out of a total of 35 students in the year, 32 used the help of an online chatbot with artificial intelligence to solve the task. The professor included a line of invisible text in the exam instructions, which was printed in the same color as the background. The hidden instruction was that the artificial intelligence should insert the word “Madagascar” somewhere in its answer in a way that made no sense. The students, who simply copied and pasted the entire text of the task into the chatbot input field without reading it, unknowingly transmitted the hidden command.

The assignment required students to compare the technological advances of the Industrial Revolution with today's times. Because no one had carefully reviewed or read the generated texts before submission, 32 students' written answers contained the word "Madagascar" in completely inappropriate contexts. As a result, they initially failed this part of the exam.

Despite the obvious evidence, the professor offered all affected students the opportunity to appeal the grading. Only two students responded to the invitation to speak, with one student being successful. She explained that because she was using dark mode on her screen, she noticed hidden text and assumed it was an actual part of the official test instructions.

While automated detection of abuse through “statistical ambushes” can sometimes lead to false accusations if students actually notice the invisible instructions or if the instructions are not entirely clear, Gibson stressed that his intention was not to publicly shame students, but to encourage real learning and critical thinking. The case comes amid a growing debate among educators about how to properly enable and regulate the use of modern technologies in the educational process.