Julie Schell’s AI Futures talk offers faculty a look at designing AI tutors around the learning challenges students face
Sean Corp, Associate Director of Communications
Fundamental to learning is what a student does when they don’t know an answer or get a question wrong. The steps they take next reveal as much about their learning as the answer itself. Recognizing a mistake, reconsidering an assumption, and trying a different approach are essential to developing understanding — and faculty are examining how AI tutors might support that process.
Julie Schell of the University of Texas at Austin will share what instructors are learning from designing and using these AI tools as student tutors during a Nov. 17 keynote at the University of Michigan. Through classroom examples, she will show how faculty can shape AI tutors around specific teaching challenges and assess whether students’ interactions encourage deeper thinking.
Part of the Center for Academic Innovation’s AI Futures speaker series, “Taking Teaching with AI Seriously: Socratic Tutors as AI Pedagogy” will take place from 1-2 p.m. in the Michigan Union’s Rogel Ballroom.
“For faculty considering how AI might support learning in their courses, seeing where colleagues began can provide a valuable starting point,” said James DeVaney, founding executive director of the center and associate vice provost of academic innovation. “AI is a big topic, so it is critical that we make room for people to question assumptions, examine evidence, and learn from one another. Julie is a dynamic, thoughtful scholar and speaker whose cross-disciplinary work gives us a concrete way to consider how thoughtfully designed AI can help students work through difficulty, reflect on their thinking, and become more independent learners.”
See the teaching choices behind the technology
Schell will share case studies of faculty across disciplines who have designed and tested Socratic AI tutors in their courses. Attendees will see the challenges instructors wanted to address, the prompts and instructions they developed, and what they learned from students’ experiences.
Schell, UT Austin’s assistant vice provost of academic technology and director of the Office of Academic Technology, conceived UT Sage, a platform that allows instructors to create AI tutors tailored to their courses. UT Sage uses a guided design process that asks faculty about their students, learning goals, common misconceptions, and planned activities. Instructors can also upload course materials that serve as the tutor’s primary source of information. Sage also integrates with Canvas, allowing students to access faculty-created tutors within their courses.
“The objective is to provide opportunities for extending students’ cognitive presence around content outside of class,” Schell said. “In less fancy terms—learning. Learning happens when you spend time thinking about and working with and applying the things that you’re learning in class, outside of class.”
Look more closely at how students learn
Drawing on a year of student-tutor interactions, Schell will examine evidence of students reflecting on their thinking, working through errors, and directing their own learning. These interactions offer faculty a way to examine what happens as students work toward an answer and where they may need additional support.
“As faculty test new approaches, we have an opportunity to ask more demanding questions about what students are learning and how we know,” DeVaney said. “Sharing that evidence helps promising ideas travel across disciplines and gives our community a stronger basis for deciding where to go next. That is how individual experiments can contribute to lasting improvements in education.”
Schell’s keynote invites faculty to consider those questions in the context of their own courses, with examples that can inform both what they try and how they evaluate the results.
