Loading…
26
Unpopular opinion: AI-generated code reviews are worse than no reviews at all
I joined a team six months ago that switched to an AI tool for all our pull request reviews. At first I thought it was cool, but now I realize it just flags fake issues and misses real bugs. Last week the AI told me to add error handling on a function that already had it, then ignored a null pointer problem my human reviewer caught in 30 seconds. The team spends more time shutting down false alerts than actually improving code. We used to have real discussions about design and edge cases, now it's just clicking 'dismiss' on AI suggestions. I bet if we tracked time wasted, it would be over 10 hours a week across our five devs. Has anyone else ditched AI code review tools and gone back to humans?
2 comments
Log in to join the discussion
Log In2 Comments
laurar383mo ago
That "arguing with a bot" bit really hits home, I've been there too.
2