When Yaniv Brandvain began teaching his biometry and biostatistics course a decade ago, he didn’t expect it to change much. He assumed statistics was a static subject and his approach would be, too. It proved far more dynamic. From the integration of R software to the shift away from doing calculations to understanding concepts, what and how he teaches has changed significantly. Now, another new variable has emerged: artificial intelligence.
Brandvain is embracing the new. Case in point, his new biostatistics textbook, a free web-based publication, provides students with guided learning experiences. At the end of each chapter, students find links to customized AI tutors designed to help them understand specific concepts, ask questions, and receive instant feedback outside class.
An associate professor in the Department of Plant and Microbial Biology, Brandvain was inspired to create the new book for several reasons. The potential of AI to enhance learning was among them.
Brandvain cited a study demonstrating that students using a custom AI tutor achieved significantly higher learning gains in less time compared to traditional active learning methods. This convinced him that AI could serve as an effective learning tool.
“I think it's because you can let your curiosity drive things more,” he says. “Figuring out how to use it as an active learning tool is something I think we need to think about."
The AI tutors in the book were created using detailed prompts containing the chapter's learning goals, the course's specific notation, and Brandvain’s teaching approach. This ensures students receive answers and guidance consistent with how the material was taught.
Samantha Pelletier, a second-year Ph.D. student in the Plant and Microbial Biology Graduate Program, took the course to better understand the right questions to ask when analyzing data. “Stats is intimidating to me and I knew this class was going to be pretty large, so the idea of having a personalized tutor was exciting,” she says.
Pelletier used the AI tutor to talk through difficult concepts. “I often made the AI use examples like relating the concepts to Taylor Swift or other pop culture ideas to make the ideas more accessible,” she says. “It was also helpful with studying, as it made practice exam questions that were trained on the real exam. … I think it helped me know quickly what I didn't understand!”
Adelaide Hazen, a second-year Ph.D. student in the same graduate program notes: “I think the fact that it was trained by the book allowed for very concise answers to confusing themes presented. The tutor was especially helpful for preparing for exams in that it was able to present practice questions based on all of the material and explain concepts when I got practice questions wrong. I do think that it really helped my learning outcomes as turning to other AIs or websites might have caused more confusion by using different definitions.”
While not all students are enthusiastic about using artificial intelligence, Brandvain thinks the benefits are real. Students who used the AI tutors performed better on tests.
He sees a future in which textbooks as we know them may give way to guided, dynamic AI-powered learning experiences. Brandvain says: "In future, textbooks could become fully AI, guided by parameters that a professor sets.” -Stephanie Xenos