DX Research analyzed over 400 organizations and found that AI-authored code has jumped from 27.4% in Q1 to 51.9% in Q2 2026, consistent across organization sizes. Alongside this, median PR size nearly doubled from 44 to 72 lines over the past year. The growth in PR size raises concerns about code complexity, review bottlenecks, cognitive debt, and maintainability as AI-generated code tends to be more verbose than human-written code. The findings prompt engineering leaders to reassess whether review and quality processes have kept pace with AI-driven output volume.
Table of contents
AI-authored code is consistent across organization sizesMedian pull request size has nearly doubled3K Impressions1 Comment