An Empirical Study on the Empowerment of Criminal Law Case-Based Teaching in Public Security Academic Institutions by Generative Artificial Intelligence

Mao Dan

Institute of Criminal Investigation, Ningxia Police Vocational College

Abstract：

Leveraging a university-level teaching reform project, this study constructs a tripartite teaching model—comprising an "intelligent case repository, a specialized prompt system, and a closed-loop process spanning pre-, during, and post-class phases"—and conducts two pilot trials. A quasi-experimental design was employed, assigning 150 pilot students to the experimental group and previous cohorts of comparable academic level to the control group, with evaluation metrics including academic performance, higher-order legal competencies, learning satisfaction, and overall teaching effectiveness. The data indicate that pilot classes achieved significant improvements in pass rates, excellence rates, case analysis score rates, compliance rates with legal documentation standards, and student satisfaction levels. This study further analyzes practical risks associated with large language models—such as hallucinations, cognitive inertia, and data security—while proposing corresponding mitigation strategies. The findings demonstrate that integrating general-purpose large language models with existing intelligent platforms can enhance the quality of criminal law case teaching at low cost, offering a replicable practical solution for public security and political-legal academies.

Key Words：

generative Artificial Intelligence; police academies; criminal law; case-based teaching; OBE philosophy; empirical research