Abstract: Based on cross-group surveys and multi-source forum discussions, this study systematically analyzes the profound impacts of large models on higher education in computer science. The research reveals that large models are accelerating the transformation of educational paradigms from “knowledge transmission” to “competency development” by restructuring teaching logics, driving innovations in curriculum systems, and reshaping core competency frameworks. However, urgent challenges need to be addressed, including inadequate industry-academia collaboration, insufficient AI skill readiness among faculty, and weak awareness of ethical risks. This article proposes a “Quad-element Education Ecosystem” framework and a dynamic response mechanism, thereby providing theoretical support and practical pathways for the transformation of higher education in the AI-native era.