From the Illusion of Learning to Doing Science: Active Learning, AI, and Authentic Research

Louis Deslauriers

Despite overwhelming evidence that students learn more when actively engaged, traditional lecturing still dominates university STEM classrooms. Why do students and faculty continue to accept the traditional lecture as effective instruction? Drawing on our Science 2011 study, I will illustrate what high-learning-density active learning looks like—many decisions and feedback cycles per unit time—and why it is so effective. I will then discuss findings from our PNAS 2019 paper on the illusion of learning: when instruction is more effortful and thus perceived as disfluent, students often feel they learn less even as their performance improves. This “fluency bias” favors polished lectures over productive struggle and can distort student judgments of teaching, leading to systematic bias in course evaluations against the most effective forms of teaching. I’ll also share how we’re bringing AI into the classroom to enhance active learning. In one application, we use real-time AI to simultaneously answer all student questions during class, increasing engagement and interaction. In another application, we’re developing a high-performance AI tutor that mimics expert human teachers, offering personalized support at scale. I’ll then conclude with examples of authentic undergraduate research as an advanced form of active learning. Several recent projects at Harvard, including one in collaboration with Michal Zajaček here at Masaryk University, have shown that when undergraduates work on well-scoped, publishable research problems, they can make genuine contributions to science while gaining a deep, lasting understanding of how scientific thinking actually works.