Stanford University instructors are using AI tools to help students find answers on their own. The work was shared through a series of talks hosted by the school’s Center for Teaching and Learning (CTL).
The “Teaching with AI Community Share-out” series lets instructors present how they have used AI in their courses. It is part of AI Meets Education at Stanford, an effort run by the Office of the Vice Provost for Undergraduate Education.
Over the past two academic years, 23 instructors have presented. They came from schools including Engineering, Medicine, Education, and Humanities and Sciences, as well as Stanford Online High School.
“The intent of this event series is not to judge the projects or spotlight the speakers,” said Kenji Ikemoto, associate director for academic technology at CTL. He said the goal is to build a community around AI in teaching. CTL plans to keep hosting these events.
Tools That Help Instructors and Students Plan
Amit Kaushal, a clinical associate professor in the School of Medicine, built a real-time question system for lectures. The tool uses AI to group, rank, and summarize student questions.
Kaushal said it helps him see when several students are confused about the same thing. It also lets students ask questions anonymously. People have since used the tool at town halls and conferences around the world.
Waheeda Khalfan, a senior lecturer in biology, helped build an AI pipeline that studies years of course content. It identifies what each foundational biology course teaches and what students should already know.
The team used the results to create self-check tests for the first week of class, paired with study guides. They also built the Foundations Navigator, a web tool that helps students choose the order of their courses.
“Something that would have taken months, we’ve now reduced to weeks,” Khalfan said. She added that humans closely review the work and the process will never be fully automated.
AI Tutors That Ask Questions Instead of Giving Answers
Shima Salehi, an assistant professor of education, worked with chemistry lecturer Jennifer Schwartz Poehlmann to design an AI tutor called STEPS. It walks students through complex science problems without giving them the answer.
Salehi said students were frustrated at first. Over time, they came to value that the tool teaches the problem-solving process. Some said it “sounds like Dr. Schwartz coaching us.”
Ariel Stilerman, an assistant professor of East Asian languages and cultures, introduced BungoBot in his winter graduate seminar. The tool acts as a tutor for classical Japanese, asking questions instead of handing out translations.
Stilerman said he wanted the seminar open to students from other majors. By week seven or eight, students had shaped the tool so much that “it was their tool as much as it was mine,” he said.
Students later deciphered scrolls with poems by tea master Ōtagaki Rengetsu, which Stanford recently acquired.
In the spring quarter, Stilerman and two PhD students introduced two more tools. Hiki is a reverse dictionary for writing classical prose and poems. Sata is a chat partner that acts like an 11th-century court lady.
With these tools, students moved fast enough to use spare class time to transcribe undeciphered manuscripts for the Stanford library.
