Many members of WashU’s faculty remain not fully aligned over technology use and collaboration and the development of artificial intelligence (AI). After the launch of the provost office’s +AI academic initiative to integrate AI into the classroom, divisions continue to ensue concerning AI and opinions on the University’s largely positive position on the technology.

Vice Dean of Undergraduate Affairs in the College of Arts & Sciences Erin McGlothlin said that she has been aware of the development of an AI-forward learning and teaching program for over a year. As part of the Undergraduate Education Advisory Committee, McGlothlin has worked to design a set of AI principles for Arts & Sciences undergraduate students and faculty to follow, which were sent out to instructors on Monday and will be circulated to Arts & Sciences students at the beginning of the fall semester. 

“The WashU Center for Teaching and Learning has been helping faculty think through AI and teaching, but there hasn’t been a coordinated effort within Arts & Sciences until now, and that coincides, I think, with a greater desire on the part of both faculty and students to have guidance regarding AI,” McGlothlin said. The principles emphasize student learning as the key objective. 

Because college historically functioned as a means to achieve career opportunities, students have been taught that product is more important than process. This perspective may incentivize cheating and AI usage, McGlothlin said. In order to prevent wrongful uses of AI, professors need to reemphasize the importance of learning for the sake of learning to students. 

However, in the future, McGlothlin thinks that students who “master AI” rather than “letting AI master them” will be more prepared. 

“In 20 years, the person who has advanced thinking skills will be more successful in their career. We cannot allow AI to make all our decisions for us; rather, we have to develop critical thinking and communication skills. So the person who hasn’t cultivated the necessary cognitive capacities is going to be lower level, just doing the inputting, outputting stuff,” McGlothlin said.

Assistant Professor and Founding Director of the Forced Migration Initiative at the Brown School Mitra Naseh, views the emergence of AI as comparable to previous technological advances. 

“When Google search started, at the time, there were academics against it. They believed that students will never use libraries again, or they will miss the opportunity for critical thinking, or research skills will vanish. But the truth is that people adjust, that they learn how to use Google or other search engines as a tool to do better in terms of academic work and research,” Naseh said. 

In her “New Media, New Technologies” course this semester, Associate Professor of Art History and Archaeology Ila Sheren has used AI as a framing device for conversations regarding learning and ethics, trying to explain to her students the importance of doing the work themselves. 

“I don’t ban AI in the classroom because it’s impossible to enforce, and I feel like if you have a rule that you can’t enforce, you’re setting yourself and the students up for failure.”

Sheren has noticed that most of her students, especially those in Sam Fox School, do not need much convincing, as they are worried about the future of AI in art and what it means for their careers. 

“I have tons of com des [communication design] majors in this class, and … they already see AI taking their career prospects and stealing work from other artists,” Sheren said. 

Sheren personally has had her work stolen by AI companies to train their large language models (LLMs), and is part of the Anthropic lawsuit, in which the authors of books pirated by Anthropic are suing for the nonconsensual use of their work to train Anthropic’s Claude AI models. McGlothlin’s books have also been stolen and input into the models, and she is also part of the lawsuit. 

In addition to dealing with AI as a professor, McGlothlin has also seen and felt that AI has damaged the perceived importance of academic research.

“We became professors because we believe in this intellectual enterprise, and it’s an attack on us. In fact, it feels like an attack on my very identity that the writing skills I spent decades developing have become completely downgraded and no longer seen as important,” McGlothlin said.

Teaching Professor in College Writing Eileen G’Sell has mixed feelings about the “generalized positive tenor at the University,” especially for first-year students, which she primarily teaches.

“The reason why I am ambivalent is not because I am against AI in every application. To be so would be a bit naive at this point. It’s crystal clear to me that across disciplines, AI can be incredibly useful to aspects of research. That being said, one of my big concerns is that, with the overall positive vibe from the university, students are getting really mixed messages,” G’Sell said.

Naseh also believes that students are confused by the range of AI acceptance between professors.

“This is a message to my colleagues: I hope that they can find a common ground to be unified in front, because I think we are also making the students very confused right now with different approaches and we are conveying different value messages. If one of us appreciates AI, one of us advocates against it, it is just a very confusing environment that students need to navigate,” Naseh said.

Over the past two years, G’Sell has noticed a decline in the long-text writing and reading abilities of incoming students. This change worries G’Sell and has motivated her to change her AI policy from some allowance with brainstorming to a complete usage ban. 

“I never expected to become this kind of teacher. In the past I was the ‘quirky culture critic’ teacher that really embraced the role of writing in online forums, and encouraged creative fervor in the class,” G’Sell, who teaches a Navigating New Media writing class, said. “Because of how rampant LLM reliance has become, I’ve instead had to adopt really strict policies. If I don’t, students have such competing definitions of what is allowed.”

While G’Sell has become stricter with AI policies, she has also become more lenient on grammatical, punctuation, and mechanical error, which she would in the past docked points for.

“I don’t want students to have any excuses for not writing in their own voice,” G’Sell said.

In order to best utilize AI tools, G’Sell thinks that students first must learn how to read and write at a high level on their own. G’Sell has noticed that, far too often, some students view AI not as a collaborator but rather an authority figure.

“Obviously, this approach is dangerous for so many reasons. In terms of faculty-student interactions, some students address faculty as though we are similar to Claude. It can feel uncanny — I’m a human being, not an obsequious responder to prompts,” G’Sell said.

While Sheren understands that AI is a necessary tool for some fields and careers, she does not view the university-wide AI initiatives as beneficial. 

“I have an engineering student in my class right now who is absolutely against the use of AI in a class like ours and probably would supplement their coding with [AI] in their own classes. I would leave it to the STEM faculty to explain exactly how it’s useful in those contexts, but no, I don’t think applying a universal standard at the University makes any sense,” Sheren said.  “The +AI initiative, at least from faculty that I talk with in my own circle, it feels as if it’s not really helpful and it feels like we’re just going to endorse AI all the time and not really think about which disciplines do need to push back against its use.”

Professor of Spanish and Second Language Acquisition Joe Barcroft points out similarities between language processing systems in humans and AI when opportunities present themselves, he said. He also gave an example of how AI might be used to promote second language acquisition provided that some important considerations are addressed.

“Presenting words in a second language using multiple voices or using multiple talkers, this kind of acoustic variability we’ve demonstrated can have really strong positive effects on second language vocabulary learning. So what a nice opportunity to be able to use AI, ask it ‘Create me six different voice types, or six different talkers,’’’ Barcroft said. “The issue is, who are those talkers? How is AI getting its information? Who are the larger stakeholders that are making the decisions about how that information is being generated? And so forth.”

When it comes to using AI to learn content, however, Barcroft is cautious about other types of pitfalls. 

“I think instructors are very protective of their students. And the last thing we would want is to guide you to use this tool and then it leads you to erroneous information,” Barcroft said.

In fact, the deadline for the summer cohort training on faculty learning AI has been extended, possibly due to a lack of signups. Sheren said that summers for faculty are meant to be a time to rest, a time to do research, and the “carrot” for joining the AI training cohort has not been enticing enough for professors like Sheren.

Barcroft completed the AI Canvas course and felt that it did a good job defining ethical uses of AI and could serve as a good introduction to AI in education for professors that might be more skeptical about AI.

“I think it would be worthwhile to do the course, and they’ll have maybe even a greater appreciation for what they’re doing and not doing as it relates to AI,” Barcroft said.

McGlothlin has chosen to offer an AI option for the final project of her Ampersand course, “Representations of the Holocaust in Literature and Film.” There are two parts to the project — a reflection essay and a presentation plan or another form of teaching about Holocaust representation. For the second component, students can choose to have AI draft a presentation script and then add to it and write a critique of the initial output. 

Naseh does not fault students for using AI on assignments since it is a tool easily at their disposal now and will be in future situations. Instead, she has her students complete activities that are not transferable to AI, such as home visits for refugees in the area. 

“My colleagues might hate me for this, but if [the] outcome of a class is an essay that can be generated by AI, I will question the value of that course. We need to update our courses to be able to bring value beyond something that can be generated by AI.”

Regardless of how faculty navigate AI in individual classes, student learning should remain centered in professors’ minds, McGlothlin said.

“If AI’s not used to further student learning, then it’s not being used effectively,” McGlothlin said. “So, it shouldn’t be that we pivot to AI just because that’s what everybody’s doing, or pivot to AI because the tools are shiny, or pivot to AI because it’s easier. It should always be: What are we doing to develop students’ cognitive skills, their abilities, their knowledge?”