AI Is Writing Code, But Who Is Reviewing It?

AI is changing frontend development, creating new challenges like code review fatigue and forcing teams to rethink junior developer roles and architecture.

A panel of engineering leaders at an InfoQ presentation explored the profound impact of artificial intelligence on frontend and mobile development. The discussion, featuring insights from Luca Mezzalira, James Hall, Danielle An, and Ivan Zarea, moved beyond simple productivity gains to address the complex challenges emerging as AI becomes a core part of the software engineering workflow. Key topics included the changing role of junior developers, the rise of "vibe coding," the very real problem of AI-generated code review fatigue, and the need for new architectural strategies. The experts warned that while AI accelerates code production, it also introduces significant risks to code quality, team dynamics, and the long-term health of a codebase if not managed proactively.

At the heart of this shift are new development patterns and their consequences. The panel described "vibe coding" as a process where developers provide high-level, descriptive prompts to AI tools rather than writing explicit, line-by-line logic. While this can speed up initial creation, it often produces large volumes of code that still require careful human oversight. This leads directly to "code review fatigue," a state of exhaustion from sifting through hundreds or thousands of lines of AI-generated code. Human reviewers must not only check for correctness but also ensure the code is maintainable, secure, and aligns with the project's architecture. The sheer quantity and often subtle, non-obvious errors generated by AI models make this a far more draining task than reviewing human-written code, creating a new bottleneck in the development pipeline.

The integration of AI into development teams is part of a broader industry trend toward automation and efficiency, mirroring past shifts like the adoption of automated testing and continuous integration. Just as those technologies changed the role of quality assurance engineers, AI is redefining what it means to be a developer. The panel suggested that the value of a junior developer is shifting away from their ability to write boilerplate code—a task now easily handled by AI—and toward their ability to think critically, design systems, and effectively prompt and guide AI tools. This creates a challenge for engineering managers, who must now devise new training and career paths for entry-level talent. The discussion also touched on the economic ripple effects, particularly for open-source software, as the value proposition for certain projects may change in an AI-driven world.

To navigate this new landscape, the panelists advised a strategic, not just tactical, approach. Engineering leaders must establish strong "architectural guardrails"—automated checks, linters, and strict coding standards—to prevent AI from introducing low-quality or inconsistent code. Code review processes need to evolve, potentially using AI-powered tools to pre-screen submissions or focusing human attention on high-level logic and design rather than syntax. Looking ahead, the experts urged teams to begin designing systems that are not just built with AI but are built for AI. This means creating architectures that autonomous AI agents can understand and safely modify in the future. Ultimately, successfully leveraging AI in frontend development requires treating it as a powerful but imperfect tool that demands new forms of human oversight, strategic planning, and a fundamental rethinking of team roles and responsibilities.

For developers and engineering leaders, the rise of AI in coding isn't just about productivity. It fundamentally changes team structure, the nature of code reviews, and the skills required for junior roles, demanding new strategies for maintaining code quality and architectural integrity.

Companies embracing AI for development face a trade-off between speed and quality. Without new processes for managing AI-generated code, businesses risk accumulating technical debt, experiencing review bottlenecks, and seeing the skill development of their junior engineering talent stagnate.