← sources

How is your org handling AI peer-review?

source: reddit (reddit:ExperiencedDevs)published 2026-11-18 14:40ingested 2026-11-19 02:36

retained text

When a developer creates a pull request that's mostly AI-generated, how do you handle peer reviews?

At my org, pull requests are mostly AI generated. Most of my peers handle peer reviews by putting the pull request back into AI.

I was doing the same for a while, but now I carefully review every single PR regardless of the source. It's not fun, it's not pretty, and it takes me a long time, but until the org says that human PRs are not necessary, that's what I'm going to do.

I leave a LOT of comments - I'm hoping this will shed light on the *actual* code quality of AI-generated code. Some teams aren't using agentic instructions, so maybe a PR or three littered with comments will encourage them. Even the fully-curated agentic code still suffers from quality issues, even from frontier models, just not as many.

I assume some orgs have simply abandoned peer-reviews, while others encourage it, and even those are probably being waved through most of the time (or just put back through AI).

What's your experience?

document fetched 2026-11-19 00:26 · canonical https://reddit.com/r/ExperiencedDevs/comments/1xzguyq/how_is_your_org_handling_ai_peerreview