Rebecca Thornton, an economics professor in her second year at Notre Dame, still remembers a student who came to her upset. It was the student’s senior year; another professor had run her history essay through an AI detector, and the software had flagged it. She needed that class to graduate.

“I said, ‘Did you do this?’” recalled Thornton. “And she said, ‘Yes, I used AI for this.’”

Thornton wasn’t shocked. The student didn’t care much about the assignment, she said, so it was easy to see why she took the shortcut.

That student has plenty of company. A Lumina Foundation-Gallup survey released in April found 57% of U.S. college students use AI for coursework at least weekly, while about half say their schools discourage or prohibit it.

Notre Dame leaves AI policy up to each professor, an approach Thornton said makes her proud. She and Thomas Mustillo, an associate professor of politics, have used that freedom in very different ways.

Thornton is generally against students using AI. She compares AI-written papers to cheap flat-pack furniture.

“It looks like a pretty glossy veneer,” she said. “But if you push too hard on it ... you’ll destroy the table.”

In the writing course Thornton has taught for 20 years, students produce an original empirical paper. Last year, she set up a Google Doc for each student to watch their writing develop over the semester. When a draft looked AI-generated, she called the student in to talk about it.

Part of her objection is also personal. Thornton said she dislikes reading AI writing and would rather spend her time responding to students than to a robot.

Meanwhile, Mustillo has brought AI into his classroom, though it took time. At first, he said, the uncertainty around AI left him paralyzed. What moved him was an interview New York Times columnist Ezra Klein did with an executive at Anthropic, the company behind the chatbot Claude. The executive argued that what students need most is good taste and good judgment, two things AI can’t supply.

“I need to teach the students how to use the tool,” Mustillo said. “And then I have to teach them how to use good judgment.”

In his classes, that means working the way political scientists increasingly do. Researchers have long read party manifestos, where political parties lay out what they stand for, line by line and sorted the contents into categories. Journals in Mustillo’s field now publish guides on handing that first pass to AI.

His students do the same, then have to defend the results. Each one picks a party and a country and explains why. They sketch out how they plan to work and draw early conclusions, and Mustillo questions them at every step. If a conclusion doesn’t hold up, he sends the student back to read the manifesto.

“That’s my goal,” he said. “To get them to think about it.”

He admits the hardest part is still unsolved.

“How do you separate what the machine did from what the student did?” he said. “That’s the problem.”

That line gets harder to hold the more students rely on AI. A study of more than 95,000 undergraduates published in Science in May found 26% of daily AI users admitted cheating with it, compared with 7% of monthly users.

The two professors also read history differently. Thornton pointed to the telegraph, the telephone and the internet, each hailed as world-changing. Those technologies did change things, she said, but people kept coming back to the basics.

Mustillo sees AI as the next step in that same line of change, and a bigger one.

“There was the analog age, the digital age and now we’re entering a new age, the AI age,” he said. “I think it’s going to be transformative.”

Where the two professors meet is motivation. Students are less tempted to take shortcuts, Mustillo said, when the work means something to them. Thornton said it’s on her to make the material interesting enough that students don’t feel the need for AI.

“When my child lies, you can blame the child for being a liar,” she said, “or you can say, who am I that that boy feels the need to lie to me?”

Their goals overlap more than their policies. Mustillo said a liberal arts education will matter more as computers take over technical skills, because people will need to bring judgment and values.

Thornton described the university she loves as a place “where you can do things slowly and think deeply.”