Generative AI has rapidly penetrated the fields of architecture and urban design, leading to an inversion of Bloom’s cognitive taxonomy. Students can quickly generate design schemes with AI, yet struggle to master the physical principles that underpin design rationality, resulting in a competence gap between creation and comprehension. This study proposed the "Inverted Bloom" C A U R M teaching model tailored for the AI era, which structures teaching in five sequential stages: Co Create, Analyze, Understand, Refine, and Memorize. Based on two courses at Tongji University—Urban Physics and Building Physics (English)—a series of teaching activities was developed on envelope thermal design, thermal comfort diagnosis, street-canyon microclimate analysis, urban soundscape design, and other topics. Empirical data were collected through classroom observation, assignment analysis, and learning feedback. Practice demonstrates that the C-A-U-R-M teaching model takes creation as its point of entry and progressively advances toward higher-order creative work through spiral iteration. In this process, AI is repositioned from a tool for task substitution to a partner for collaborative co-creation and an object of critical reflection. The model significantly enhances students’ classroom engagement, deepens their understanding of core concepts such as heat transfer mechanisms, thermal comfort, the urban heat island effect, and soundscape, and encourages them to evaluate design solutions through physical reasoning and, in turn, distill personalized rules of thumb for design practice. The model emphasizes that AI should serve as an auxiliary tool for physical reasoning and judgment rather than replace students’ creative thinking. It therefore provides a transferable pedagogical framework and practical pathway for architectural technology courses in the AI era to move beyond passive adaptation toward proactive curricular and pedagogical reconstruction.