This research investigates the application of immersive teaching in Civics, enhanced by knowledge distillation techniques. We employed feature distillation and sample-balanced distillation algorithms to refine both the teacher and student networks, ensuring effective content delivery and deeper knowledge acquisition. Experimental results demonstrate that the student networks, optimized with distillation and identification loss, significantly outperform their unoptimized counterparts in feature retention. Furthermore, the integration of IDA data enhancement and kvDKD decoupled knowledge distillation methods substantially boosts the model’s efficiency and accuracy. Our teaching experiments revealed heightened student engagement and learning outcomes in Civics, with all students covering 8 of 12 topics and an average of 27 students scoring above 95 per topic. A comparative analysis with a control group confirms the efficacy of immersive teaching approaches in improving student performance and the quality of Civics education.