The prevalence of generative artificial intelligence tools, including Copilot and ChatGPT, has brought brand-new transformation directions for coding education within university computer majors. Traditional undergraduate coding courses are trapped by rigid unified teaching arrangements, insufficient personalized tutoring support, obsolete workshop projects, and oversimplified assessment standards, which cannot cultivate engineering talents matching industrial recruitment standards. Based on years of frontline teaching practice in programming modules, this paper reorganizes the core dilemmas of conventional coding pedagogy, explores how generative artificial intelligence improves classroom teaching efficiency, sorts out four complete teaching links covering pre-class preparation, in-class lectures, offline workshops, and post-course evaluation, and puts forward targeted management measures to address student over-reliance on intelligent tools. The research can provide practical reform references for the digital upgrading of university programming courses.