Educators outline do's and don'ts as students turn to AI chatbots for schoolwork

Three years after ChatGPT’s debut, students at schools and universities across the United States are routinely turning to AI chatbots for homework help, prompting educators to set guidelines about what constitutes legitimate use and what crosses into academic dishonesty.

About two dozen states have issued state-level AI guidance for schools, though application remains uneven, according to the Associated Press. Individual institutions have adopted policies ranging from blanket restrictions to no universal rule at all.

Educators broadly agree that AI can serve as a legitimate study aid when it supplements a student’s original work. Using chatbot output in place of one’s own thinking, they say, undermines learning and may violate academic integrity policies.

Don’t copy and paste

The most basic line educators draw is that passing off chatbot output as one’s own work is off-limits.

“AI can help you understand concepts or generate ideas, but it should never replace your own thinking and effort,” the University of Chicago says in its guidance on using generative AI. “Always produce original work, and use AI tools for guidance and clarity, not for doing the work for you.”

Yale University’s Poorvu Center for Teaching and Learning warns that the consequences extend beyond ethics: “If you use an AI chatbot to write for you — whether explanations, summaries, topic ideas, or even initial outlines — you will learn less and perform more poorly on subsequent exams and attempts to use that knowledge.”

Do use AI as a study tool

Educators say AI performs well in a tutor or study-buddy role.

Casey Cuny, a California high school English teacher, advises students to upload class notes, study guides, and course materials to ChatGPT, then prompt it to: “Quiz me one question at a time based on all the material cited, and after that create a teaching plan for everything I got wrong.”

Cuny posts a traffic light on a classroom screen to communicate permitted uses. Green-lighted activities include brainstorming, asking for feedback on a presentation, and research. Red-lighted, or prohibited, uses include asking an AI tool to write a thesis statement, a rough draft, or to revise an essay.

Sohan Choudhury, CEO of Flint, an AI-powered education platform, recommends using ChatGPT’s voice dictation function to narrate one’s understanding of a subject. “I’ll just brain dump exactly what I get, what I don’t get” about a subject, he said. “I can go on a ramble for five minutes about exactly what I do and don’t understand about a topic. I can throw random analogies at it, and I know it’s going to be able to give me something back to me tailored based on that.”

Check your school’s policy

Policies vary widely across institutions. The University of Toronto’s stance is that “students are not allowed to use generative AI in a course unless the instructor explicitly permits it,” directing students to check course descriptions for specific guidance. The State University of New York at Buffalo, by contrast, “has no universal policy,” leaving decisions on AI tools to individual instructors’ academic freedom.

Be transparent with teachers

Choudhury said the climate around AI in classrooms has shifted since ChatGPT’s 2022 launch. “Two years ago, many teachers were just blanket against it. Like, don’t bring AI up in this class at all, period, end of story,” he said. Three years on, “many teachers understand that the kids are using it. So they’re much more open to having a conversation as opposed to setting a blanket policy.”

Rebekah Fitzsimmons, chair of the AI faculty advising committee at Carnegie Mellon University’s Heinz College of Information Systems and Public Policy, said students frequently do not recognize when they have crossed a line.

“Often, students don’t realize when they’re crossing a line between a tool that is helping them fix content that they’ve created and when it is generating content for them,” Fitzsimmons said. She helped draft new AI guidelines for Carnegie Mellon students and faculty aimed at creating greater clarity.

Cite AI use

The University of Chicago says students should acknowledge AI if it was used to generate ideas, summarize texts, or help with drafting. “Just as you would cite a book or a website, giving credit to AI where applicable helps maintain transparency,” the university says.

Oxford University requires AI tools to be used “responsibly and ethically” and in line with its academic standards. “You should always use AI tools with integrity, honesty, and transparency, and maintain a critical approach to using any output generated by these tools,” the university says.