Hacker News Reader

At best this would be 1.7x more _discovered_ bugs. The average PR (IMO) is hardly checked. AI could have 10x as many real issues on PRs, but we're just bad at reviewing PRs.

View on HN

It's been suggested that code reviews shouldnt be about finding bugs, because they do not really work well for that goal.

https://www.microsoft.com/en-us/research/wp-content/uploads/...

Couldn't this be true in the other direction as well? Anecdotally I see developers putting a lot more scrutiny into vibe coded PRs, while AI code tends to be highly commented (by the AI) and potentially easier to read.

I've seen way more human comments of "I don't know what this does but if I remove it everything breaks" in systems.

Reply on news.ycombinator.com