In recent years, the educational landscape has struggled to keep pace with generative AI. Educators, unprepared for these swift advancements, face the challenge of navigating a spectrum between two extremes: from avoiding AI altogether to thoughtfully integrating it in their pedagogical practices. Educational research now has the crucial task of providing practical, discipline-specific strategies to use AI effectively. Computer science is particularly impacted by the integration of AI. Now embedded directly within many software development suites, these tools are transforming professional coding practices, enhancing efficiency and quality. However, they also impact how students learn to code, presenting both challenges and opportunities. This shift calls for an adaptation in teaching practices with a dual focus: preparing students to use AI tools thoughtfully and critically, while reinforcing core programming skills. On both fronts, cultivating metacognitive skills is essential to ensure that generative AI serves to enhance rather than replace fundamental coding abilities. This approach envisions a learning environment with automation complementing critical and hands-on coding. This paper makes a twofold contribution. First, it proposes an educational intervention that aims to help students critically assess AI in programming. This intervention emphasizes critical thinking, enabling students to navigate AI tools with discernment. Second, we introduce a novel comprehensive, multimodal data collection methodology to analyze how students interact with generative AI in coding contexts. This approach provides a robust foundation for evaluating student-AI interactions and refining future pedagogical interventions. The preliminary findings provide encouraging information for educators and improve the role of AI in computer science education.