Observable Changes in Open Source Development Patterns Before and After AI Coding Tools: A Multi-Repository Study

Background: AI-powered coding assistants, including GitHub Copilot and ChatGPT, have seen widespread adoption since late 2022. While controlled studies demonstrate individual productivity gains, the aggregate impact on open source ecosystems remains unclear. Objective: We investigate observable changes in open source development patterns before and after the widespread adoption of AI coding tools. Method: We collected monthly metrics from 21 major repositories across 9 technology clusters (Python web, JavaScript frontend, systems languages, AI/ML-native, and others) spanning January 2020 to December 2025 (72 months). We compare pre-breakpoint (before November 2022) and post-breakpoint periods using Mann-Whitney U tests and Cohen's d effect sizes. Results: Of 315 statistical tests (15 metrics × 21 repos), 156 showed significant changes (α=0.05). The strongest signal: issue close time decreased substantially (mean effect d=-1.08, "large" by convention). 19/21 repositories showed increased AI-related mentions in commits and discussions—a proxy signal for tool awareness and adoption. However, commit velocity and contributor counts showed negligible changes on average. Conclusions: Development patterns measurably changed coinciding with AI tool adoption. Faster issue resolution may indicate efficiency gains, though we cannot establish causation. The lack of change in commit velocity challenges simplistic productivity narratives. Replication: Interactive dashboard at https://ai-oss-impact-study.kotaicode.solutions. Code and data at https://github.com/kotaicode/ai-oss-impact.