Harvard Study Finds AI Coding Agents Slowed PR Merges by 49% Despite Higher Output
A working paper from Harvard economists Fiona Chen and James Stratton analyzed data from 718 firms, covering 49.1 million pull requests between 2021 and 2026. Their findings indicate that coding agents substantially increased code and PR volume but did not meaningfully raise completed Jira issues. Average merge times rose by 49%, and the share of PRs needing formal change requests nearly doubled. The authors attribute the bottleneck to review capacity, which the article argues cannot be solved by having one LLM review another's code. It proposes a layered verification approach that combines deterministic checks with AI review tools.
This is an AI-generated summary. ShortSingh links to the original source for the complete article.



Discussion (0)
Log in to join the discussion and vote.
Log in