Accenture’s GitHubCopilot Study
summary
Accenture runs a six month controlled study across 450+ Accenture developers using GitHub Copilot and compares them to a similar control group, finding developers accept about 35% of Copilot suggestions and retain ~88% of accepted code after edits, attempt more builds with a 45% higher build success rate, complete ~15% more pull requests, and report 100% usefulness with 50% saying it is extremely useful.
AI-written summary · about
chapters (6)
- 0:00Introduction to the study
- 1:07Accenture engineering overview
- 2:11The case for AI in development
- 3:58Early experiences and adoption
- 5:05Research findings and metrics
- 7:26Developer feedback and future
related talks
50mHow GitHub operationalizes AI for team collaboration and productivity
40mUnlocking Productivity with GitHub Copilot Best Practices and Tips
01h 00mLevel up with GitHub Copilot: using AI to learn, code, and build
34mLessons Learned from Building GitHub Copilot(s)
31mGithub Co-Pilot: Addictive or Effective?
50mGitHub Copilot: Coding Will Never Be The Same Again
26mWhen Agile and AI Collide
45mAgentic AI for Maintainers: Getting the Most From GitHub Copilot Enterprise