PR Size Over Time (Greptile, 2025)
Median lines changed per PR, March–November 2025
Lines of Code per Developer (Greptile, 2025)
Monthly lines of code per developer, March–November 2025
AI-Generated Code Share Over Time
Estimated % of new code that is AI-generated (U.S. / global)
GitHub Copilot User Growth
Cumulative GitHub Copilot users (millions)
AI Coding: Key Metrics
| Metric | Value | Source |
|---|---|---|
| Developers using AI tools (2025) | 84% | Stack Overflow Survey |
| Copilot code suggestion acceptance rate | ~30% | GitHub / ACM |
| Code generated by Copilot per user | 46% | GitHub |
| Task completion speedup (Copilot) | 55% faster | GitHub controlled experiment |
| Experienced dev speedup (METR RCT) | 19% slower | METR (2025) |
| Self-reported time savings | 3.6 hrs/week | DX (135K+ developers) |
| PR size increase (Mar–Nov 2025) | +33% | Greptile |
| Lines per developer increase | +76% | Greptile |
| AI code market size (2025) | $7.37B | Industry reports |
| Copilot market share | 42% | Industry reports |
| Bugs per developer with AI adoption | +9% | GitClear |
| Developers who don't fully trust AI code | 96% | Sonar |
Sources
Greptile: The State of AI Coding 2025 2025
Analysis of ~1 billion lines of code/month. Tracks PR size, lines per developer, SDK downloads, and AI tool adoption from March to November 2025.
METR: AI Impact on Developer Productivity 2025
Randomized controlled trial with 16 experienced open-source developers on 246 tasks. Found AI tools made experienced developers 19% slower despite feeling 20% faster.
GitHub: Copilot Productivity Research 2024
Controlled experiments showing 55% faster task completion with Copilot. Tracks acceptance rates, code retention, and developer satisfaction across 4,800+ developers.
Stack Overflow Developer Survey 2025: AI 2025
Annual survey covering AI tool adoption, trust, and usage patterns. 84% of developers use or plan to use AI coding tools.
GitHub Octoverse 2024 2024
Annual report on GitHub platform activity. Tracks Copilot adoption, user growth, and the impact of AI on developer workflows and repository activity.
GitClear: AI Code Quality Research 2025
Analysis of 211 million changed lines of code. Tracks code churn, copy/paste rates, and refactoring trends in the AI era. Found 41% of code is AI-generated and a 4x increase in code clones.
Qodo: State of AI Code Quality 2025
Report on AI-generated code quality, trust, and adoption patterns. Covers developer sentiment and code review challenges.
DX: Developer Productivity Analysis 2025
Analysis across 135,000+ developers showing 3.6 hours/week average time savings from AI tools. 22% of merged code is AI-authored.
Faros AI: The AI Productivity Paradox 2025
Research report exploring the gap between perceived and measured AI productivity gains. Teams with high AI adoption merge 98% more PRs but PR review time increases 91%.