jazzband
Update PR template with question about using an LLM?
I would like to add a checkbox to our PR template that indicates if the change was at least partially generated with an LLM/AI and points to a LLM/AI policy that communicates our stance.
From my perspective, LLMs are tools that can help. However, they tend to be verbose and generate lots of code. If I'm reviewing that code, I want to be able to respect the changes to the level the human put effort into them. If I know a person spent 10 hours crafting a PR to generate a change, I'll be more accepting of minor differences. However, if it's been generated with an LLM/AI with less effort, I think it's fair to be a little more aggressive on asking for changes. After all, the LLM/AI isn't going to care and will gladly make those changes in a few minutes.
The challenge is, it's hard to know when this is the case or not. I think editing the PR template may help new contributors know that a LLM disclosure is appropriate.
[Edit]: I wrote this as a statement, but it's definitely a question open to discussion.
As a person who grudgingly started using LLMs too I agree.
I think this is a good post: https://xeiaso.net/notes/2025/assisted-by-footer/ I'm not sure requiring such a footer would be a good idea (because the tooling doesn't really make it easy).
