A Michigan State University professor says the AI detection tools some instructors use are not reliable enough to prove a student cheated. Daniel Kramer, who teaches at the school’s James Madison College, told The State News he treats detection results as a reason to talk with students, not as proof.
His comments come as students at the East Lansing school raise concerns about being falsely accused of using artificial intelligence. Some say the fear has changed how they feel about turning in their work.
“I don’t believe they’re reliable enough to say that you necessarily used AI, or you didn’t use AI,” Kramer said. He has used detection tools before but does not rely on them alone.
Students Worry About False Flags
Economics freshman Poe Meh said she has seen students on social media describe being wrongly accused of using AI. She said those posts have made her nervous about submitting assignments.
“It’s something that I’m scared of because I’ve seen how some students get accused of using AI when they didn’t,” Meh said. She added that being accused after putting in hard work “doesn’t make you feel good.”
International relations junior Matthew Horne also questioned the tools. “You’re relying on AI to tell you that I used AI when I actually didn’t use AI,” he said.
Horne said he uses AI to come up with ideas and to get started on assignments. He said students should still be responsible for their own work. “It’s what your brain is supposed to be used for, not the computer,” he said.
Kramer allows some AI use in his courses, including editing, spell-checking and refining writing. He generally does not allow AI to produce first or final drafts.
Oral Exams and Handwritten Tests Considered
Kramer suggested oral exams as one option, along with conversations where students explain or defend their papers. He said students who leaned too heavily on AI would likely struggle to show a strong grasp of their own work.
He said James Madison College already focuses on critical thinking, writing and working with several sources. The college also uses in-person blue book exams, which make it harder for students to rely fully on AI.
Kramer said oral exams and extra meetings would take a lot of time for professors and students. That challenge is greater in larger classes, he said.
Horne said some of these ideas could work in other programs but should fit each field. He gave the example of computer programming courses using handwritten exams to test understanding.
Meh said handwritten assignments could help but are not practical for every course. She also noted that some professors use AI as a teaching tool in their lessons.
She said changing formats would not stop all cheating. “To be honest, even if we do it on paper, students who want to cheat will find a way,” Meh said.
Meh said colleges should focus on building trust between students and professors. “I think we just need to have more trust in our students,” she said.
Kramer’s survey of students in one of his courses found that attitudes toward AI differ across MSU colleges. He said professors must find “that fine line” between allowing some AI use in class and limiting students’ learning.