We know that college students are using AI tools to help them research projects and write papers. But the problem apparently is much worse than we might have imagined. For 34 years, Robert Serrano has taught economics at Brown University. In one of his classes, he typically draws 30 students that produce midterm averages between 65 and 80 percent.

When there was a tragic shooting on campus, he decided to give the students a take-home exam. Enrollment nearly tripled to 86. Then the grades came back. Nearly half scored perfect or near-perfect grades. He and his assistants noticed the student responses matched ChatGPT phraseology. 

He told the class that the final would be in-person and closed-book. If the scores were similar, he would count both. If not, the midterm grade would be dropped. More than two dozen (27) dropped the course immediately. Among the ones who stayed, the final exam average dropped to 48 percent. Many (19) of the students failed.

He had proved his point. But there is more. The professor submitted the evidence to Brown’s Standing Committee on Academic Code. He heard nothing. Therefore, he went public with his findings. The press picked up the story, and it went viral.

I realize this is only one story, but I suspect it could be repeated hundreds of times in the classes of the thousands of universities in this country. Now that these AI tools are available, students are letting these computers do the thinking for them. They aren’t learning. They are punching numbers into their laptops and asking questions of chatbots.

I am encouraged that this professor did what he could to stem the tide of AI cheating. We need professors and colleges to deal with this problem.