Entry

Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity

Joel Becker, Nate Rush, Elizabeth Barnes, David Rein

Synopsis

RCT with 16 experienced open-source developers across 246 tasks; AI access increased completion time by 19%, contradicting developer forecasts of 24% speedup before and 20% after — a sharp metacognitive miscalibration.

Keywords

·METR ·developer productivity ·RCT ·Cursor ·slowdown

Open paper ↗ arXiv ↗ Report issue ↗

Related entries

Enhancing School Students' Self-Regulated Learning through Generative AI Support: A Randomized Controlled Trial

2026 · Educational Psychology Review 2026
AI Assistance Reduces Persistence and Hurts Independent Performance

April 6, 2026 · arXiv
How AI Impacts Skill Formation

January 28, 2026 · arXiv
Evolving with AI: A Longitudinal Analysis of Developer Logs

January 15, 2026 · arXiv
How AI and Human Behaviors Shape Psychosocial Effects of Extended Chatbot Use: A Longitudinal Randomized Controlled Study

March 21, 2025 · arXiv
Generative AI Without Guardrails Can Harm Learning

July 2024 · PNAS 2025