Ambera Forge: AI-authored code often costs less review
TL;DR. Ambera Forge data shows AI-authored code often requires less review time than human-written code across thousands of open-source repositories. - Many repositories see attributed AI work reviewed at equal or lower rates compared to human contributions. - The 'AI review tax' is not universal; 71% of projects pay the same or less review on AI-generated code. - Methodology analyzed 5,388 repositories and over 444,000 merged pull requests for attribution.
- Ambera Forge analyzed 5,388 open-source repositories and 444,225 merged pull requests.
- Data indicates 71% of projects pay the same or less review for AI-authored code compared to human code.
- The median repository shows AI code costing less review, challenging the 'AI review tax' assumption.
- Only 29% of comparable repositories experience higher review costs for attributed AI work.
Sources
- The median open-source repo pays no "AI review tax" (data from 5,388 repos) — forge.ambera.app