In 'I Skipped Physics Class and AI Made Me Regret It', published in Canadian Teacher Magazine's Fall 2026 issue on September 20, 2026, I introduce Anchored Recalibration Learning, a method for AI-assisted, self-directed learning of complex topics. The learner starts from a real anchor, something curious or confusing they can see or touch, and then uses conversation with an AI model to adjust the difficulty until the idea sticks. The idea grew from a visit to a science museum with my children, where a spinning-chair exhibit on gyroscopic precession, torque and angular momentum led to an hour-long AI conversation, and from an animated clip about two avatars reaching a banana, which I use to ask when learners need speed, depth or both. The article offers teachers and students a five-step guideline: find something confusing, ask for a visual, push back on dense answers, form and test hypotheses and generalize to new situations. I also argue that teachers should set conditions, time and rewards rather than force learning, and that not every problem deserves the slow path. Read the full article at https://canadianteachermagazine.com/2026/09/20/i-skipped-physics-class-and-ai-made-me-regret-it/ (by Marvin Starominski-Uehara )