When Coding Stops Being the Scarce Skill

Generative AI is changing what it means to learn and demonstrate expertise in programming education. As code generation becomes increasingly automated, educational emphasis shifts from code production toward understanding, evaluation, critical mediation, and professional judgment. This paper examines these changes and proposes a four-layer model comprising Foundational Understanding, Model-Building through Construction, Critical Mediation, and Professional Judgment. The model provides a framework for reconsidering learning activities and assessment in AI-mediated programming education, while maintaining human responsibility for evaluating, justifying, and validating AI-supported solutions.