SOUTH CHARLESTON, W.Va. –West Virginia University at Parkersburg assistant biology professor Joel Farkas discussed the impact of artificial intelligence on students and faculty during a recent state Higher Education Policy Commission meeting.
Farkas, who also serves as chair of the state Advisory Council of Faculty, delivered a report to the commission on how students are using artificial intelligence in their coursework.
He said early versions of artificial intelligence were primarily used to quickly retrieve facts and answers from the internet. He noted that newer models can perform complex tasks, including reasoning, mathematics and genetics.
“It used to make mistakes in problems, and I could predictably beat ChatGPT and then every year, systematically everything I beat it on it automatically figures out how to beat me back,” Farkas said.
Farkas said he specifically explains to students how artificial intelligence should and should not be used. He said he encourages students to use AI to generate resources and receive quick back-and-forth feedback.
He said artificial intelligence can be a valuable tool when used properly, though some students continue to misuse it.
“It’s at the point now where you can’t trust the authenticity of anything that students do unless you watch them do it in the classroom without a computer,” Farkas said. “I know there are some professors going back to all pencil and paper, in some cases literally pencil and paper, because they can’t trust that students are doing their homework outside of class without the use of AI.”
He cited an example involving a math professor at WVU Parkersburg, saying she adjusted her teaching approach after students began using artificial intelligence to complete homework assignments.
Farkas said he observed a shift over the past three years, from students averaging 60-70% on both homework and tests to situations where some earned 100% on homework but only 40% on tests. He explained that the professor has since adopted a model in which students’ complete readings to prepare for class and work on problems during class time.
He said teachers will likely need to rethink how they approach instruction in response to artificial intelligence.
“I think a lot of us are probably going to rethink our teaching in that kind of way where I think, were going to have to go back to that more or less that flipped model where every assessment is done in class,” Farkas said. “Or we have to rethink our grading scheme because a lot of use have developed a grading scheme based on what students have done for the past 10 or 20 years, but its changed drastically.”
He said online classes are the most affected, as students can complete coursework using artificial intelligence. He suggested that online tests may need to be proctored or time-limited to prevent students from looking up answers.
Farkas also said institutions might need to restrict credit transferability, noting that students could take a challenging course online elsewhere and transfer the credit back.
“I’ve seen a lot of instances where somebody has one class they specifically want to avoid, and so they take that one class online somewhere else and transfer back,” he said.
He gave the example of students at WVU-Parkersburg who take courses such as anatomy and macroeconomics online at other institutions because they find them too difficult, then transfer the credits back.
He said he would like to see more institutions adopt formal artificial intelligence policies.
He said WVU-Parkersburg allows departments to implement AI policies, but not all have done so.
“Most people that I’ve talked to at my school, don’t have any kind of statement on AI, you just kind of informally talk in the classroom about what you should do and what you shouldn’t,” Farkas said. “I would love to see some top-down guidance on that.”