Reconstruction and Practice in agent-driven AI practical teaching

To address the practical challenges in current AI training courses, such as the disconnect between training content and industry demands, the lack of authenticity in practical scenarios, and the limited diversity of evaluation methods—this study takes the curriculum reform of the ”AI Engineering Training” course at our university as a starting point to explore the development of a new agent-driven model for AI training. This model, driven by both industrial needs and technological innovation, restructures the practical training system into a three-tier cognitive progression: foundation building, system integration, and cutting-edge innovation. It establishes a four-dimensional collaborative education mechanism that integrates ideological and political education, blended teaching, industry-research feedback, and diversified evaluation. The entire teaching process is built around the development of intelligent agents. The teaching reform bridges the gap between talent cultivation and industry demands, significantly enhancing students’ engineering skills and innovative capabilities. Accordingly, it outlines a replicable pathway for AI practical teaching reform in higher education institutions.