How to review AI code without reading the code
A Faros AI report shows a 76.3% increase in pull requests merged without human review since April, with many companies automating 50–80% of code reviews using AI tools. While AI excels at mechanical code inspection, the shift eliminates scheduled team conversations about system architecture and decision-making. The author argues that code review serves a dual purpose: catching bugs and building shared mental models of the codebase, and that replacing human review entirely sacrifices the latter without solving the accountability gap AI tools cannot bridge.
Key points
- PR merges without human review jumped 76.3% since April.
- AI handles mechanical inspection well but lacks accountability for design decisions.
- Teams lose critical discussion moments about system architecture.
- Hybrid approach needed; AI alone creates 'review theater' without substance.
Summary and key points written by AI from the headline and a short excerpt, so they can be incomplete or wrong. The New Stack did the reporting; read the original for the full story.
Read the full story at The New Stack
