From Passive to Participatory by Leveraging Artificial Intelligence for Active Learning Environments

Abstract

The rapid integration of Artificial Intelligence (AI) into educational practice offers unprecedented opportunities to transform classroom pedagogy from passive, lecture-centered approaches to participatory, learner-driven experiences. This study reports the design, implementation, and evaluation of an AIenabled active learning framework through a quasi-experimental study involving two matched student sections for the secondsemester B.Tech Computer Engineering course Python Programming at RK University, involving 120 students. The intervention blended AI-assisted pair programming, adaptive lowstakes quizzing with real-time feedback, AI-driven Socratic tutoring for conceptual clarity, and analytics-informed instructional adjustments, all within an explicit ethical AI use policy. A quasi-experimental design was employed, with two matched sections: an AI-Active group incorporating AI tools into active learning strategies, and a Traditional-Active group relying on established active learning methods without AI integration. Comparative analysis demonstrated that the AI-Active cohort achieved higher final exam scores (+7.8 points), improved lab task accuracy (+11 percentage points), reduced programming anxiety, and shortened time-to-solution, while also exhibiting increased engagement in formative assessments. These outcomes align with recent findings from published work indicating moderate-to-large effect sizes for AI-enhanced instruction, particularly when sustained over multiple weeks and supported by structured guidance. The study concludes that embedding AI into active learning can enhance both cognitive and affective learning outcomes in programming education, offering a scalable model for modern classrooms. Recommendations for sustaining gains, ensuring academic integrity, and scaling the approach across technical disciplines are provided. However, limited research has compared AI-enabled active learning directly with traditional active learning in large programming cohorts.