Rethinking Code Review in the Age of Generative AI: A Conceptual Framework for Layered Software Verification

Abstract

Generative artificial intelligence (AI) is transforming software development by increasing the speed and volume of code production, while simultaneously challenging assumptions underlying conventional code review. This study examines how generative AI creates structural mismatches in software review and proposes a layered verification architecture for AI-assisted development. The study adopted a qualitative conceptual research approach using an integrative review design. Data were collected from scholarly literature, conference publications, and selected industry evidence on code review, AI-generated code, AI-assisted review, and human-AI collaboration. The literature was analysed thematically around four issues: volume, authorship, failure modes, and accountability. Findings indicate that AI-assisted development can increase review workload, weaken the assumed relationship between authorship and understanding, introduce plausible but semantically incorrect code, and complicate independent human accountability. The study proposes a four-layer verification architecture comprising deterministic automated verification, AI-assisted contextual verification, independent human technical review, and human accountability and governance. The study concludes that generative AI does not eliminate the need for human code review but requires a redesign of its role. It recommends risk-based review, complementary use of automated and AI tools, and preservation of independent human judgement. The study contributes a structured framework for adapting software verification to AI-assisted development.