Course-Specific AI Chatbots Piloted in 2 Expos 20 Courses

Two faculty members teaching Expository Writing 20, a required first-year class run by Harvard’s Writing Program, have piloted the limited use of artificial intelligence chatbots in the curriculum.

Expos 20 courses have been required since 1872, but with the advent of generative artificial intelligence tools, the classic course is now adapting.

James Herron, a preceptor for Expos 20: “The Ruling Class,” has designed a course-specific chatbot — which has been trained on the class’s readings — to help students in “refining their argument, making it more persuasive.”

“What they’re asked to do then is to enter their thesis into the chatbot,” Herron said. “What the chatbot does is it supplies four plausible counter arguments to their thesis.”

“I’ve scripted most of what it says to the students,” Herron said. “So the part where the AI is really doing its thing is where it’s producing the possible counter arguments to the students.”

Jane Rosenzweig, a preceptor for Expos 20: “To What Problem is ChatGPT the Solution,” has also piloted a similar chatbot.

“The chatbot really does want to write your paper for you, and so we’re trying to strike a balance,” she said.

Rosenzweig added that expository writing classes are not the only courses piloting these customized chatbots.

“There’s some really interesting pilots going on in other parts of Harvard,” she said. “The Physics department has piloted a chatbot for the big intro Physics class. And I think it’s going to turn out that they’re useful in different ways in different departments.”

Students said they have mixed feelings about the integration of AI in their expository writing classes.

Isaac B. Hertenstein ’28, a student in Herron’s class, said using ChatGPT to ideate is “not the best idea.”

“I just think it’s inaccurate and also kind of takes away from the humanity of the writing and what your ideas are and how you can effectively argue them,” he said.

However, Hertenstein added that he sees the merits of AI later in the writing stage when “you already have an argument or counterargument created.”

Cleo N. Carney ’28, another student in Herron’s course, said she believes that permitting the use of AI in the classroom authorizes what many students were already doing.

“At the end of the day, everyone is using it — whether they’re saying they’re using it or not,” she said. “So I think now that the cat’s out of the box, there’s no point in Harvard or any school saying ‘Okay, well you can’t use it whatsoever.’”

Herron said that many of his students “are more fluent with these technologies” than he is and he appreciates student input. He and Rosenzweig said they are still experimenting with these “relatively new” technologies.

“We’re trying to learn how to use them in ways that are productive and helpful, and it's still a work in progress,” he added.