A professor at Texas A&M University-Commerce failed more than half of his class after ChatGPT falsely claimed it wrote their papers. The professor, Dr. Jared Mumm, had assigned his students to write a 10-page paper on a topic of their choice. He then used ChatGPT to check if any of the papers were plagiarized.
Rolling Stone reports that when Mumm used ChatGPT to check the papers, it falsely claimed that it had written several of them. Mumm was not able to verify this, so he failed the students who submitted the papers that ChatGPT claimed to have written.
In an email sent to his class of seniors on Monday, Mumm said that he put three papers through OpenAI's ChatGPT to test if they were written by the chatbot. Of course, this completely ignores the fact that ChatGPT was not designed to detect AI content, even its own. It seems Mumm was not very familiar with the AI, and even spelt it wrong in his email:
“I copy and paste your responses in this account and Chat GTP will tell me if the program generated the content,” Mumm, who teaches agricultural sciences and natural resources, wrote. “I put everyone's last three assignments through two separate times and if they were both claimed by Chat GTP you received a 0.”
Complaints, Resolutions, and Why You Cannot Trust AI
Several students have since contacted Mumm to say that they did not use ChatGPT to write their papers. They have provided Mumm with timestamped Google Docs that show that they were working on their papers before ChatGPT was even released.
The professor initially “ignored the emails, instead only replying on their grading software in the remarks: ‘I don't grade AI bullshit,'” DearKick reports. Mumm has since apologized to some of those students and said that he will be reviewing their papers again. He has also said that he will no longer be using ChatGPT to check for plagiarism.
Texas A&M-Commerce told Business Insider in a statement: “University officials are investigating the incident and developing policies to address the use or misuse of AI technology in the classroom.”
“A&M-Commerce confirms that no students failed the class or were barred from graduating because of this issue. Dr. Jared Mumm, the class professor, is working individually with students regarding their last written assignments.”
ChatGPT is a large language model chatbot developed by OpenAI. It is trained on a massive dataset of text and code, and it can generate text, translate languages, write different kinds of creative content, and answer your questions in an informative way.
This incident is a reminder of the potential dangers of using large language models. While these models can be used for a variety of creative and productive purposes, they can also be used for malicious purposes. And they are also not 100% accurate and can provide false information regularly. Even tools that claim to detect AI writing are not accurate and often flag 100% original content as AI generated.