Harvard researchers find AI coding agents shifted the bottleneck to code review
Summary
ComplexDiscovery reports on a Harvard working paper by Fiona Chen and James Stratton showing that AI coding agents accelerate draft code but push the bottleneck to code review, with review time rising around 49% per pull request across nearly 726k workers in 718 firms. The piece highlights how AI-assisted drafting interacts with human review, governance, and potential pricing implications for software development and DevOps workflows.