LLM reviewers are useful, but some PR checks should stay deterministic

With AI coding agents increasingly adept at creating pull requests, the nature of code reviews is evolving. While these large language models (LLMs) can assist in reviewing code, they introduce new challenges, such as ensuring that changes align with the intended task and verifying the repeatability of evidence for merging. The shift means that some traditional deterministic checks in PR reviews remain crucial to maintain code integrity and design coherence. This evolving landscape highlights the need for a balanced approach, combining the efficiencies of AI with the reliability of human oversight.