The recent Provost’s Seminar on Teaching explored the opportunities and challenges generative AI is bringing to graduate education at the University of Michigan.
The event, co-sponsored by the Provost’s Office and the Center for Research on Learning and Teaching, or CRLT, brought nearly 150 faculty, staff and academic leaders to the Michigan Union on May 13.
In his welcoming remarks, Matt Kaplan, associate vice provost, learning and teaching, and executive director, CRLT, said that the day’s programming would be a continuation of a conversation that began last November with the Provost’s Seminar on GenAI and undergraduate education — but it would tackle the topic through the unique lens of a U-M graduate student.
“As we planned today’s event and talked to faculty and staff around the campus about it, it became clear that this is in many ways a more complex topic [than genAI in undergraduate education], given the multiple roles played by graduate students,” Kaplan said, citing the range of degrees grad students pursue and the fact that many assume multiple roles, including student, instructor and researcher.
Laurie McCauley, provost and executive vice president for academic affairs, also spoke briefly at the program’s outset, reflecting on the question: “Which parts of graduate education really matter, and why?”
“A moment like this requires more than a policy conversation, although we know policy matters. It requires more than a technology conversation, although technical understanding also matters,” McCauley said. “It requires us to ask what kind of intellectual habits graduate education is meant to cultivate.”
“We know our pedagogy must protect authentic human thinking against the temptation of effortless automation. We’re preparing students to use generative AI responsibly, but we’re also preparing them to remain intellectually alive in a world where answers are abundant and attention is fragile,” she said.
An expert panel, moderated by Michael Solomon, vice provost for graduate studies and dean of the Rackham Graduate School, examined how higher education has navigated moments of consequential change in the past, revealed insights from an AI working group of graduate students, and discussed how U-M might help guide and support its grad students at this moment.
The panelists included:
- Ann Austin, University Distinguished Professor of Higher, Adult, and Lifelong Education, Michigan State University
- John Carson, associate professor of history, LSA
- Keanu Heydari, Ph.D. candidate in history, LSA
- Nigel Melville, associate professor of technology and operations, Stephen M. Ross School of Business
Austin kicked off the discussion by acknowledging that this time in higher education is “not business as usual.”
“As we talk about AI, we need to be thinking about the implications for many aspects of higher education as we look forward,” Austin said, drawing parallels to the transformative period in higher education that occurred after World War II, when there was a sharp increase in student enrollment and an influx of federal money.
“We know that skepticism is greater for higher education. We know that the polarization in our country contributes to different points of view about higher education,” she said.
“We’re also dealing with … the rather dramatic change in the relationship between higher education and the federal government. The partnership that was established post-World War II is certainly under some challenges.”
“My point,” Austin said, “is that there are a variety of major changes occurring in higher education that have implications for how we prepare our graduate students… in both promising and challenging ways with the AI developments.”
Heydari, who served on an AI working group with other graduate students, shared insights from those discussions, including how students are actually engaging with AI.
“The education and research group’s interviews showed that writing assistance was cited by under 20% of users. Ideation and execution were each cited by over 50% — and by ideation, they meant brainstorming and literature work, and by execution, they meant coding and data analysis,” he said, adding that another group said it used AI primarily as an organizer or synthesizer that helped them decide “what to do next.”
Heydari also said the working groups revealed that a significant number of both students and instructors were unaware of U-M’s central GenAI resources page, and that students are entering this AI transition with a wide range of preparedness, calling the disparity an “uneven baseline” that complicates an already complex issue.
Melville addressed the application of AI in the workplace and said he believes “we all have a responsibility to develop an informed and educated point of view about AI. This does not mean we are pro this or con that … those are ethical judgments. But I believe our responsibility is to be informed and educated, just like we do with everything else.”
He also pointed out that when it comes to being informed about GenAI, graduate students often know more than their professors — and that it’s important that this new period of exploration be collaborative, and that faculty approach it with humility.
“I often start my graduate classes or my discussions with my doctoral students by saying, ‘Look, I know a lot about AI, but I don’t know everything, and you’re going to know things that I don’t know,” he said. “‘So, we’re working together here, and we’re going to have a time in each class, in each conversation, where we’re each going to share some things.’”
Carson referenced Charles Dickens’ “It was the best of times, it was the worst of times,” to describe the current GenAI environment, and he pointed to how some AI opportunities, like improved efficiency, can have both positive and negative impacts.
“Sometimes efficiency takes away from human beings, from spending the time doing the work and going places never anticipated … because we want the answer, a quick answer. And right now, we’re in the world of demanding quicker and quicker answers,” he said.
“But I think collectively, we as educators, we in universities, need to start learning how to push back on that and say, ‘Things take time. You don’t need to do this as fast as you think you do.’”
Following the panel, four concurrent roundtable discussions tackled how GenAI may impact the different roles graduate students assume at U-M:
- Graduate students as learners
- Graduate students as teachers
- Graduate students in research and scholarship
- Graduate students in experiential, creative or clinical settings
During the roundtable sessions, CRLT consultants facilitated discussions and took notes on the suggestions, concerns, challenges, and opportunities identified by faculty — then these observations were shared on posters attendees could browse during lunch. The event’s lunchtime session also included an interactive conversation with Dean Solomon, with faculty sharing highlights from their roundtable discussions.
In the coming days, CRLT will post a video of the panel discussion and share with attendees and university leaders a summary of the roundtable conversations.
