In a University of Illinois introductory data science course, professors discovered widespread cheating on attendance checks and received AI-generated apologies from students. The incident, involving over 1,000 students, highlights tensions in higher education amid AI use. Instructors addressed the class on October 17, treating it as a life lesson without disciplinary action.
The Data Science Discovery course at the University of Illinois, taught by statistics professor Karle Flanagan and computer scientist Wade Fagen-Ulmschneider, enrolls more than 1,000 students across multiple sections. Attendance and participation contribute a small portion to the grade, tracked via the Data Science Clicker tool. Students scan a QR code in class to access a personalized multiple-choice question, with a 90-second window to respond.
A few weeks into the fall semester, Flanagan and Fagen-Ulmschneider noticed more responses than students physically present. They investigated by checking site refreshes, device IP addresses, and server logs, revealing that classmates were alerting absent students when questions appeared.
The professors contacted around 100 suspected students, issuing warnings and requesting explanations. Responses included a flood of apologies, which initially seemed contrite. However, about 80 percent were nearly identical, featuring phrases like "sincerely apologize," indicating AI generation.
On October 17, during class, the instructors displayed a mash-up image of the apologies to highlight the issue. "We reached out to them with a warning, and asked them, ‘Please explain what you just did,'" Fagen-Ulmschneider said in an Instagram video. Flanagan described the episode as a hoped-for "life lesson." No formal discipline followed, and the matter was handled lightly as a caution against such practices.
University of Illinois subreddit discussions from teaching assistants and students underscore broader AI misuse in the course. One TA noted students using AI for easy problems, often employing untaught functions that exposed the cheating. Another claimed AI appeared in 75 percent of submitted work, including weekly 75-word paragraphs. Concerns also arose about false accusations, given the unreliability of AI detection tools.