Under the intelligent transformation of architectural design courses in higher education, three major bottlenecks constrain teaching quality: outdated textbooks, inadequate teaching methods, and fragmented tools and resources. Based on Bloom's Taxonomy of Educational Objectives, this study constructed a three-stage competency progression model of "cognition-application-innovation", integrated cognitive load theory, designed and implemented a lightweight AI teaching module, and embedded it into the core course series "Architectural Design 3-5" in an architectural design program at a university for three consecutive semesters. Specifically, the empirical study used "Architectural Design 4" as the experimental scenario with a pre-test and post-test control group design. The results showed that the lightweight AI module can effectively reduce teachers' preparation difficulty and teaching time without affecting the original course structure, while significantly enhancing students' design ability and higher-order innovative thinking. The research shows that embedding the lightweight AI module and reconstructing core teaching scenarios and tasks effectively solves the triple problem of "textbooks, teaching methods, and tools", providing a replicable practical path for the intelligent transformation of architectural education.