Professors called for more nuanced conversations surrounding the use of artificial intelligence in higher education at a webinar Wednesday afternoon. While the panelists acknowledged the potential downsides of generative AI and the unreliability of detection software, they also agree that AI is important in research.

The Academic Senate and the Office of the Provost jointly organized the webinar to stimulate an interdisciplinary conversation about AI and its different uses in classroom settings. Stephen Aguilar, an assistant professor of education, said communication between professors about AI is essential to understand the full extent of its use in higher education.

Daily headlines, sent straight to your inbox.

Subscribe to our newsletter to keep up with the latest at and around USC.

“We all have different perspectives, and it’s important for faculty across USC to really engage with one another,” Aguilar said. “It becomes very easy to focus on your own particular context, and not think broadly. And it’s only by thinking broadly that we can make decisions at the institutional level.”

Jen Sopchockchai Bankard, an associate professor of writing, said that many of her students were unsure if using AI counted as plagiarism and therefore afraid to use it in the classroom. She said the University should foster more nuanced conversations with students regarding when and how AI should be used. 

”Students are getting very mixed messages,” Bankard said. “Their professional contexts are encouraging them to use AI … but at the same time, it’s being disciplined in their classrooms … It’s important to create a conversation around AI in the context of academic integrity … What constitutes academic dishonesty? And how does our ability to detect it influence our policies on it?” 

Panelists also discussed what they believe to be the threats generative AI posed to their fields of work. Bodour Salhia, an associate professor and interim chair of the Department of Translational Genomics at the Keck School of Medicine, said AI could breach data privacy violations and expose patients’ private information. Aguilar said a core challenge he faced was determining how to evaluate student learning when parts of their work were AI-generated. 

Bankard said her largest concern was faulty AI detection software, which occasionally flagged the translation software international students used as plagiarism. She said the detection software has incorrectly flagged Grammarly in the past.

“There’s lots of issues in terms of bias towards English language learners in that provision,” Bankard said. “Currently, the detection part is really iffy. And so is that going to inadvertently create inequity among students who are being accused of plagiarism?”

Salhia said generative AI is also shaping clinical work and research. She said AI is improving the process of finding drug candidates, streamlining the process of creating safer, more effective drugs; accelerating clinical trials; and assisting with patient engagement, retainment and recruitment. 

“There’s a whole regulatory submission process that AI can help with. Tailoring submissions, and again, all the paperwork that’s involved, [AI] can help automate that process a little bit more and make it smarter,” Salhia said. “Which brings up the issue of worker replacement … many specialties will have to reinvent themselves to fit with the modern times of AI.”

Panelists concluded the webinar by describing the things they would like faculty to think more about in relation to AI. Bodour said faculty should think about how AI could make academia more efficient. Bankard acknowledged that AI would impact different disciplines and different teaching styles differently, but said that all professors should make an effort to take the emergence of AI as an opportunity to rethink assignments in the classroom.

“Rethinking and redefining what we’re asking students to do and what critical thinking should be, that would be my biggest suggestion,” Bankard said.