# Why don’t we write code that AI reviews instead of the other way around? — retained record

> Retained reddit record from reddit:cscareerquestions.

Canonical: https://highsignal.app/data/records/d6a544ee4a01d135

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:cscareerquestions)  published 2026-11-19 04:19 ingested 2026-11-20 02:36

[original source ↗](https://www.reddit.com/r/cscareerquestions/comments/1y0082r/why_dont_we_write_code_that_ai_reviews_instead_of/)

## retained text



```
I’m a new-ish engineer (1 YOE) at a big tech company. Of course, we’re being pushed to use AI like no tomorrow. Big outages for our service everywhere, constant crunch because tiny bugs are missed and we have to run fixes, a billion work streams and brains being stretched like putty. I hate reviewing AI code. I’ve found it’s significantly better to just write it myself, say “hey, any compilation issues or blockers?” And fix them myself or not. That way I still understand 100% of the code I write and the context around it, and my skills don’t atrophy. I don’t find that this makes me significantly slower, either, since the code AI produces usually has to go through a bunch of rounds of reviews. Why not just invert the process? 
```



document fetched 2026-11-20 00:23 · canonical https://reddit.com/r/cscareerquestions/comments/1y0082r/why_dont_we_write_code_that_ai_reviews_instead_of
