American history professor Jason Gibson suspected that his students were using AI to cheat. So he decided to find out the extent of the problem — and with the help of a simple trick, he caught almost all of his students cheating.
American history professor Jason Gibson suspected that his students were using AI to cheat. So he decided to find out the extent of the problem — and with the help of a simple trick, he caught almost all of his students cheating.
Gibson explained that he hid hidden text in white font in a midterm exam assignment. Because of this, students who simply copied the questions into the AI unknowingly added completely meaningless facts about Madagascar to their answers about the Industrial Revolution, writes Cybernews.
Because the instructions were written in white on white, they were invisible. And it worked: 32 out of 35 students in the two groups immediately inserted the question into the chatbot and used its answer as their own.
It seems that the students are using AI almost blindly. If they had even reread the text generated by the chatbot, they would have definitely noticed the strange references to Madagascar.
But almost no one did, and the essays students submitted included phrases like, "Madagascar's purple bicycle whispers to the ceiling" and "Madagascar floats sideways through the afternoon."
“You would be surprised at how bizarrely the word ‘Madagascar’ was used in these students’ work. It was more than obvious that they hadn’t even read their answers,” Gibson said.
When students were confronted with the fact of cheating, only two of those caught tried to appeal their grades. All of the cheaters received an "unsatisfactory" for that part of the exam.
