The University of Michigan is the first higher education institution to integrate AI-powered Coursera Coach, a new AI tool powered by Google Gemini, into a live online course.

Coursera Coach AI delivers personalized, interactive lessons through Socratic dialogues based on the course’s themes. The tool leverages course materials, such as videos, PDFs, articles and other files, to create customizable rubrics that ensure learners fully understand and relate the content to their experience. Learners engage in back-and-forth conversations with the AI tool, answering questions to demonstrate comprehension.

“The material really comes alive when students engage with me after the lecture or visit me during office hours,” said course instructor Vic Strecher, professor of health behavior and health equity at the U-M School of Public Health.

Strecher, who has taught at U-M for nearly 30 years, is passionate about teaching and engaging with his students. However, traditional lecturing, whether in-person or online, often felt limited in interaction. He called the integration of Coursera Coach a “dream come true” after 35 years of exploring how technology and tailored communication can enhance health outcomes.

With Coursera Coach, Strecher now feels he is teaching in three dimensions.

“When a student comes up to me with questions, I can find out more about that person and can really interact. This tool allows me to simulate that with an online course. The material can be far more interactive and engaging,” he said. “This is a dream come true for me. It’s something I’ve been interested in, literally since the 1980s.”

The AI Coach is particularly adept at recognizing when a user has satisfactorily fulfilled a prompt, highlighting relevant course concepts and redirecting engagement if responses are off-topic or negative. Strecher collaborated closely with Coursera’s team for months to fine-tune the AI-powered dialogues, ensuring they provide meaningful and helpful feedback.

All instructors on the Coursera platform can now create similar educational modules featuring Socratic dialogues using the AI-powered Coursera Coach integration.

“I’ve been working in digital health communications for 35 years,” Strecher said. “Back in the ’80s, if we could collect a little information on a person, we could use software to tailor feedback to help a person quit smoking or manage stress.”

This interest led to the establishment of U-M’s Center for Health Communication Research, which showed that personalized information yields better results.

With the advent of large language models and generative AI, Strecher now sees opportunities for tailoring and intelligent dialogue that were unimaginable in the 1980s.

“In testing, it was clear it was accomplishing just what I would want to accomplish in the classroom,” he said. “It empathizes and tries to understand where you are coming from. It evaluates the emotional words and tries to learn the underlying root cause of the response, while at the same time bringing the learner right back to the objectives of the course.”

Jeff Maggioncalda, CEO of Coursera, underscored the importance of human expertise in AI-enabled learning.

“Grounding the AI in human expertise and oversight is critical,” he said. “It is an extension of the instructor, not a replacement.”

Just hours after the AI tool went live, a Coursera employee informed Strecher that his course had already logged over 2,000 interactions between learners and the AI dialogue lessons. Strecher sees this as a significant step in reaching underserved populations efficiently.

“As a teacher and researcher, my interest in digital technologies has been to reach millions of people. And to do it efficiently and to reach people who are underserved who really need help. What makes me so excited about systems like Coursera Coach is that it allows me to do all of those things,” he said.