90% of developers use AI. Over 80% say they're more productive.
I read Google's DORA report expecting the delivery metrics to match, but they don't.
Over 5,000 professionals surveyed : throughput did improve, delivery stability got worse and burnout didn't change.
Individually, the story looks great. Faster problem-solving, better code quality, more code shipped. (Actually, I see myself in this)
Organizationally, it doesn't. AI adoption still correlates with increased delivery instability. Workplace friction: same as before despite all the productivity gains.
DORA calls it the "mirror and multiplier" effect. AI amplifies what's already there. Good engineering culture gets better. Bad processes get worse, faster.
People seem to answer "what's the ROI of our AI tools?" by using the number of PRs merged. DORA measures what matters: deploy frequency, lead time, change failure rate, recovery time. Those didn't improve.
(Self-reported productivity, not telemetry. But the instability is measured. Full report linked below.)
AI doesn't fix your engineering org. It exposes it.
Link of the report: https://cloud.google.com/resources/content/2025-dora-ai-assisted-software-development-report?hl=en®ion=US