AI Code Boom Creates Review Bottleneck

🔥 What happened Harvard researchers Fiona Chen and James Stratton analyzed 300 million work outputs from 700,000 developers. AI agents boosted code output by up to 30% — but pull request review time exploded by 49%. 💡 Why it matters The productivity gain evaporates entirely. Nearly twice as many pull requests require rework, and 14% more developers are pulled into code reviews. One senior dev put it bluntly: AI saves time on routine tasks but costs more on complex ones. ⚡ Our take Deploying AI coding tools without rethinking your review pipeline just swaps one bottleneck for a worse one. The real innovation isn't generating code — it's verifying it.