AI use explodes on Minnesota college, university campuses

Artificial intelligence is dramatically changing higher education as professors adapt to its use, despite fears and hand-wringing that college students are using it as a cheating free-for-all.

As higher education wrestles with unprecedented challenges— including shrinking budgets due to federal cuts and fundamental doubts about its value — AI’s growth is prompting instructors to have frank classroom discussions about key skills students must master before they graduate and the ethical use of tech tools.

While professors and students at Minnesota colleges and universities have varying perspectives on AI’s usefulness, many faculty are rethinking their assignments and tests. Skeptical professors are going old-school with physical test booklets and oral exams, while early adopters are boosting students’ AI use through creative projects that were impossible four years ago.

“The AI train has really sort of left the station in academia,” said Dal Liddle, an Augsburg University English professor. “These last few months are when students are really kind of jumping into it with both feet.”

The AI discussion largely began three years ago with the debut of ChatGPT, a free, online chatbot capable of realistic conversation with users. Other AI models and systems soon followed and AI use has now gone mainstream.

Now, every professor likely is pondering when and how to use AI with students, said Galin Jones, director of the University of Minnesota’s School of Statistics and chair of the Data Science and AI Hub.

“This is a significant issue that most faculty worry about: Are the students giving up their agency?” Jones said. “Not taking responsibility for learning and ... just outsourcing it to the AI agent?”

Different academic subjects use AI in different ways. AI may be applicable in a broad discipline like psychology, but less so in a small, specific field like linguistics because bots have been fed few linguistics texts, said Claire Halpert, the U’s Institute of Linguistics director. But that’s changing.

Most AI models struggle with chemistry, said Elizabeth Ness, a biochemistry major at Minnesota State University, Mankato, so she doesn’t use it often. She does use it to clarify instructions for labs or suggest articles for a literature review.

“I don’t think you can really just completely avoid it,” Ness said.

Professors said young people actually have a range of feelings about it; some oppose it for moral or environmental reasons.

“I remember one young woman saying, ‘I hate it,’” Jones said. “Other people are like, ‘Yeah, I’m using it to just help me cheat in classes that I don’t really care about.’”

“Many, many of my colleagues are ... trying to use it,” he said. “I don’t understand the appeal.”

He uses some in-class paper and pencil tests, he said, and sometimes students write in Google docs, which tracks changes, so he can look back on their process, checking for AI use. Students also submit handwritten samples so he can see how they write.

Students in Cat Saint-Croix’s philosophy classes at the U have always done their logic assignments by hand in class, but in response to AI, she also has them do nearly all exams on paper in class.

Her AI policy: students can use it to talk about ideas as they would with a peer, but not to produce work.

In the future, classes with lots of writing may schedule longer, in-class writing labs, similar to science labs. That will take class time, though, and result in covering less material, she said.

“Anybody who is using AI to grade something, I think, is making a terrible mistake,” he said.

His tests are all on paper and in person, though he knows students use AI for other things. He now assigns a final project with a spoken element. In his genetics class, students explore genetic traits in their family history through a family pedigree assignment.

“At that point it is like, ‘Why cheat on this?’” he said. “It’s going to be more effort to make up something [so personal].”

While Moumita Dasgupta, Augsburg University’s physics department chair, is a big AI fan, her students need to master physics. So she gives frequent, quick quizzes in class and decreased the weight of homework because students could use AI. She’s also included oral exams as part of midterms.

Other professors say they’re embracing AI, using it to shape assignments.

Recently, he input hundreds of product reviews for e-bikes on Amazon into ChatGPT. Then, students asked ChatGPT to explain trends in the data while also using Excel for analysis, comparing both methods.

Shilad Sen, Macalester College’s chair of the computer science department, also works as an AI scientist. His department has a “no AI” policy for intro classes, with some exceptions, he said, but students in his upper-level class learn AI skills so they can meet certain goals.

He gave them an assignment taken from his industry job: Decide how and when to use AI to rate how well a job candidate would match with several job openings and share that with the candidate. Eventually, humans will develop college courses’ content but AI will often deliver it. The classroom will be a learning community where students get feedback, he said.

At Hamline University, Wes Kisting, provost and dean of the faculty, said he hopes his school will approach AI thoughtfully, realizing that, to produce savvy AI users, “We have to address good, old-fashioned literacy.”

He cited an assignment that doesn’t let students “offload” thinking or practicing a skill to AI; a professor might use AI to generate three novel summaries, with students writing a paper critiquing them and picking the best one.

Sangupta, the physics professor, said she talks about ethics in the first days of class, telling students that cheating with AI won’t help them learn or succeed on tests. Other times, she reminds them how much they’re paying in tuition.

Macalester student Sheila Bhowmik discussed a dense text with ChatGPT for a recent assignment. It was so easy and effective, she said, that she fears her ability to resist it in a crunch.