Professors are now encouraged to establish explicit guidelines on the use and permissibility of generative artificial intelligence in their syllabi and class materials, according to an email sent by College Provost David Kotz on Aug. 30. In an interview with The Dartmouth, Kotz noted that these resources present both “opportunities and challenges,” and that the College opted to avoid implementing “blanket and broad policies.”
The announcement comes in the wake of nascent GenAI tools, such as ChatGPT and Dall-E, which produce text or other content when provided with a prompt, according to Kotz in his email.
“Most universities are where we’re at right now — allowing professors to make their own decisions about which technology is appropriate in their course and being clear about what the boundaries are, so the students know what’s allowed in that course,” Kotz said.
According to Dartmouth Center for Advancement of Learning director Scott Pauls, professors are concerned that students will use GenAI to circumvent their assignments and violate the Academic Honor Principle.
He added that detecting the use of GenAI in academic work is difficult unless the violation is “egregious,” noting that tools that claim to detect AI-generated text are ineffective and often become obsolete due to the speed at which the technology progresses.
According to Kotz, professors can best prevent Academic Honor violations through clear policies on the use of GenAI in their courses.
“The professor should be clear to the extent they can about what tech is encouraged or not, and if the students are unclear about what the professor intends, they should ask,” Kotz said. “[Students] should be submitting [their] own work and citing the work others do.”
However, Pauls added that many professors also recognize the technology will become “integrated into everything” and view it as an analog to calculators or search engines.
Pauls noted that certain professors have already asked their students to use arguments provided by ChatGPT to write essays or write critiques of ChatGPT-generated essays, which he believes are “highly metacognitive” exercises. He added that professors at the Geisel School of Medicine have created an interface that enables ChatGPT to assume the role of a patient and engage in conversation with medical students.
To encourage such innovation, Kotz said that DCAL planned to lead a working group supporting faculty interested in using GenAI in their classrooms, “so faculty can see what other faculty have been exploring.”
“I’m hoping that this working group can provide structure for faculty who want to dive in a little more deeply,” Kotz said.
Associate professor and vice-chair of the English department Aden Evens stated that the Institute for Writing and Rhetoric met to discuss the use of GenAI in the classroom. Although several professors planned to prohibit the use of GenAI tools in their classes, according to Evens, he criticized that approach as “backward thinking.”
“It seems to me that [prohibitions are] an attempt to cleave to a now outdated understanding of what writing is and how it works,” Evens said. “At this point, we use technological tools in our writing in all sorts of ways, and it’s hard to decide that GenAI is a completely different kind of tool from the word processor and Grammarly tools that have been legitimately used up until this point.”
Evens did recognize that the debate over GenAI will remain unsettled, since technological development can be unpredictable and subject to “wild, wide-eyed excitement” that leads to fear-induced reactions.
Despite the challenges in predicting the future of GenAI and establishing clear policy, Kotz said that students should avoid using the tool to shortchange their learning opportunities.
“My advice to students is that they remember the reason they’re taking the courses, and the reason they’re in college in the first place is to learn,” he said. “They should approach every course, homework assignment, exam or lab session with the goal of learning as effectively as they can.”
