The proliferation of AI-powered code generation tools challenges traditional programming education, which usually prioritizes writing code from scratch while treating code comprehension as a byproduct. As developers increasingly work with AI-generated code, however, the ability to read, evaluate, debug, and integrate code from diverse sources becomes paramount. Unfortunately, current computer science curricula lack systematic approaches for developing these receptive programming skills.
This paper presents a novel pedagogical framework that re-conceptualizes programming education through the lens of second language acquisition theory.

Drawing on the Common European Framework of Reference for Languages (CEFR), we propose six progressive stages of receptive skill development:
(1) lexical-syntactic recognition,
(2) semantic comprehension,
(3) structural analysis,
(4) pattern recognition,
(5) critical evaluation, and
(6) contextual integration.

Each stage features specific learning objectives, authentic code-reading activities, and multimodal assessments aligned with measurable competency descriptors.

Based on this framework, we introduce the concept of the “Code Curator”: a professional whose primary value lies in orchestrating, evaluating, and maintaining code quality across human and AI-generated components.

Unlike traditional programmers who focus on producing original implementations, Code Curators excel at comprehending existing codebases, identifying subtle flaws in AI-generated solutions, assessing security vulnerabilities, and making strategic decisions about component integration. This position demands advanced analytical skills, a thorough understanding of architecture, and the ability to prioritize competing quality factors like performance, maintainability, and accuracy. Code Curators act as quality control, balancing the benefits of AI-generated code with the need for system integrity and avoiding technical debt.

Our presented framework operationalizes this role through concrete pedagogical strategies informed by cognitive load theory and scaffolding principles. By initially emphasizing code reading over code writing, we reduce extraneous cognitive burden while building robust mental models of program behavior. The modular design enables integration into existing curricula without wholesale restructuring, from introductory programming courses through advanced software engineering.

This paper contributes to computing education by:
(1) providing the first systematic framework for receptive programming skill development,
(2) adapting validated language education methodologies to address emerging AI-driven challenges, and
(3) offering practical implementation guidance for educators.

As generative AI transforms software development from a production-centered to a curation-centered activity, our framework equips students with the critical evaluation skills necessary for responsible AI integration and long-term career resilience.

Our work’s impact goes beyond individual skill-building; it reframes how we understand programming expertise in our increasingly automated world, where human skills in code understanding and quality assurance remain crucial.

Keywords: AI-assisted programming, code comprehension, receptive programming skills, computer science pedagogy, pedagogical innovation.