A professor at Texas A&M University in Commerce accused students of cheating by using artificial intelligence to write essays. This was reported by The Register.
Jared Mamm, the coordinator of the Department of Agricultural Sciences and Natural Resources, checked the texts for authenticity with ChatGPT. According to him, the tool flagged some papers as machine-generated.
The situation led to a freeze on grades and a further investigation. The students under suspicion admitted using AI to write essays.
As a result, half of the academic cohort were temporarily denied diplomas.
“I’ll give everyone an X in this course. I don’t grade AI junk,” said Mamm to the students.
The university administration confirmed the use of the X grade. They said it is a temporary measure while the investigation continues.
The university noted that cheating opportunities were eliminated for some students, while others chose to submit fresh essays for assessment. At least one student admitted using ChatGPT to complete the assignment.
“A&M in Commerce confirms that no student failed classes and was denied graduation because of this issue. Officials are investigating the incident and developing policy to combat misuse of AI technologies,” the statement said.
The administration has also begun developing and implementing tools to identify AI-generated texts in coursework.
A source familiar with the situation said the dispute was nearing its end. The university allegedly acknowledged the instructor’s decision as improper.
“This was completely outside protocol and an improper use of ChatGPT. At least no one used profanity in the accusations,” he said.
According to experts, even advanced detectors of generated texts are not able to accurately identify fakes.
“As large language models improve at mimicking people, the best possible detector will achieve an accuracy of nearly 50%,” says Vinu Sankar Sadasivan, a PhD student at the University of Maryland.
He added that detectors can be deceived by paraphrasing outputs from a large language model. Sadasivan urged learning to live with the fact that we may never reliably distinguish AI text from human text.
“We should not use these detectors to render a final verdict,” he says.
In December 2022, a professor accused students of fabricating essays with the help of ChatGPT.
In January, New York state schools banned the use of the chatbot on campus. Later joined the largest universities in Australia.
In the same month, OpenAI introduced a detector for generated texts. Soon ChatGPT was taught to circumvent restrictions of the tool.