Reinventing classic courses: a paradigm reconstruction study of AI

With the rapid advancement of Artificial Intelligence (AI) technology, traditional signal processing education faces challenges such as outdated knowledge structures misaligned with industrial advancements, single teaching models and insufficient cultivation of innovation ability, and an imbalance between theory and practice, and lack of real-world application scenarios. This study adopts constructivist learning theory and backward curriculum design, proposing an “AI-empowered Four-Dimensional Reconstruction Framework”. Building on the national first-class course construction experience at Beijing Institute of Technology, this framework drives the transformation of Digital Signal Processing (DSP) curriculum from “mathematics-oriented” to “engineering problem-driven”, from “knowledge transmission” to “meta-cognitive ability cultivation”, from “tool application” to “intelligent method innovation”, and from “single assessment” to “multimodal competency evaluation.” It establishes a systematic course architecture, covering “demand analysis, content construction, teaching implementation, and dynamic evaluation”. By restructuring knowledge systems, innovating experimental methods, and reforming assessment mechanisms, a new teaching ecosystem is created, deeply integrating fundamental theories, intelligent technologies, and cutting-edge applications. Teaching practice demonstrates that the reconstructed course significantly enhances students’ innovation capabilities and complex problem-solving skills, providing a replicable model for upgrading traditional core courses in the new engineering context.