Anti-LLM Sentiment Considered Harmful

shinzui:

I’m concerned that developers who use coding agents to contribute to the ecosystem may be discouraged from participating.

I read any “anti AI” policy as “anti sloppy work” policy for the simple reason that if you’re a project maintainer you really can’t tell if someone wrote code using an LLM or their bare hands unless they added the LLM as a co-author.

If you:

• Use a library
• Find a bug
• Implement a bugfix and understand that bugfix (with or without an LLM)
• Submit a PR under your own name and take responsibility for what you’ve done

Then I find it extremely unlikely it would get rejected or labeled slop or that you would even raise any suspicion.

In the cases where you can obviously tell a PR is vibe-coded, chances are you’ve run into some very sloppy work that’s not worth the effort of even understanding (maybe that’s not the case, but who would waste their time). That’s what I understand @hasufell is talking about.