AI and Cheating: Challenges in American Universities

A crisis related to cheating is emerging in American universities as artificial intelligence (AI) becomes deeply integrated into learning. Classroom practices aimed at preventing cheating are becoming more extreme and inconsistent. False accusations against students are rising, and the definition of cheating is shifting, according to professors, students, and academic integrity experts.

Challenges in Academic Integrity

At UCLA, some students in a recent sociology class reported being instructed to obtain a mirror large enough to reflect their entire desk-area workspace and keep their laptop cameras on so the professor could monitor them during online tests. In another course, students were required to cross their arms or hold them behind their heads during oral video exams to prevent typing into AI platforms. A Los Angeles attorney representing students facing disciplinary action at California colleges noted AI-related accusations now account for about 35% of her firm’s education caseload, and the number is increasing rapidly. She cited multiple instances where professors reported AI violations by over half the class.

One significant complication is the unclear definition of cheating with varied rules about AI use. Faculty members are striving to redesign the classroom experience with some allowing AI sources in drafts, while UC Berkeley’s law school recently banned nearly all AI use. Lee Rainie, director of Elon University’s Imagining the Digital Future Center, pointed out the trust issues between students and faculty regarding AI disclosure.

Survey Findings

Igor Chirikov from UC Berkeley’s Center for Studies in Higher Education led a study on AI use among students and found that AI prevalence is growing rapidly. Of the over 95,000 students surveyed at 20 public research universities in spring 2024, around two-thirds had used AI for classwork and one-third used it regularly. About 9% admitted to using AI to cheat, like writing papers with its help when not permitted, or solving math assignments. He noted a substantial increase in AI use over the past two years but did not examine cheating, false accusations, or disciplinary practices in the current year.

University conduct officers have highlighted how plagiarism cases are now being overtaken by AI-related cases that are tougher to prove. Failed assignments suspected to be AI-generated and academic integrity hearings have led to an increase in academic AI-defense law firms. Students have devised strategies to defend or avoid discipline, especially those wrongly accused by professors or faulty detection software.

False Accusations and Student Experiences

Students who claimed they did not use AI explained being accused due to professors’ suspicions or detection software, which researchers have noted can produce false positives. Turnitin, a popular AI checker, claims its scanner is less than 1% inaccurate but acknowledges a small risk of false results. Other studies from University of Pennsylvania academics and European researchers have reported higher failure rates among common AI-scanning apps. An article from UCLA’s Humanities Technology last year cautioned that detection tools are flawed due to false positives and may rely on outdated AI technologies.

Extreme Classroom Practices

Titi Olotu, a junior at UCLA, encountered unusual AI monitoring when taking an online class. She was contemplating dropping the course following a mirror email directive. Her accommodation allowed her brief exams breaks and paper note-taking. Olotu felt online proctoring treated any movement as suspicious. Consequently, she dropped the class. The UCLA professor involved did not respond to interview requests. A school spokesperson mentioned that student concerns are taken seriously and there are processes for reviewing conflicts, but did not discuss individual cases.

While lockdown browsers and screen-sharing videos are common in online exams, mirrors and body movement restrictions are considered more extreme. Students and experts see these practices as reflecting the chaos, confusion, and fear instigated by new technology. Another UCLA sociology student, Ashley, described the degrading nature of being required to show arms and hands. Online boards are filled with student complaints over excessive policies like proctoring software tracking keystrokes, eye movements, or facial scans to detect AI prompts.

Student Defense Strategies

Students have developed their own defenses. Ivan Ornelas, a recent UC San Diego neuroscience graduate, never used AI but created every paper in Google Docs to display the version history and demonstrate his work. He also avoided including ‘AI tells’ — such as cliches and vague statements — in his writing. Aldan Creo, a master’s student at UC San Diego, took similar measures after being accused of using AI for detailed explanations in a math assignment. His grade was reduced by half until he appealed during office hours.

Adrienne Hahn, an attorney who represents students in California, highlighted that defending against AI accusations is increasingly becoming the focal point of education cases at her firm. She referred to instances where instructors report more than half of a class to academic conduct boards for AI violations. She advises students to gather evidence like texts, study notes, and version histories from Google Docs or Microsoft Word.

Tricia Bertram Gallant, UC San Diego’s academic integrity director, cautioned against using the term ‘falsely accused,’ arguing that universities may lack policies or procedures, or fail to follow them. Ardea Caviggiola Russo, the University of Notre Dame’s academic integrity office head, noted AI isn’t the only source of cheating or academic laziness. Papers still suffer from conventional plagiarism actions like nonexistent citations and fabricated sources.

Teaching Adaptations

Adam Kaiserman from College of the Canyons reported a student for AI-related cheating after his scanner identified nine instances. However, one case was overturned when the investigator’s advanced AI detection tool disagreed. Kaiserman shifted his approach to teaching, offering extra credit for students who place their phones at the front of the room before class. His humanities class assignments also changed to avoid AI-generated responses.

Gallant emphasized the need for universities to update systems to keep pace with technological advancements. She stated that it’s not about widespread false accusations but rather a system adapting to new technology.