What are the maintainers' feelings towards AI-generated pull requests?

[LucaPizzagalli]

Hi,

There has recently been some renewed interest in this repo. I think a big reason is that LLMs are now actually good at coding, so the bar for contributing to the project is much lower. But so is the value a pull request brings (anyone with access to an LLM API can replicate it).

So what is the maintainers' attitude towards AI-generated code? Do you want to keep tight control over what the codebase looks like and carefully review new contributions, or "embrace" a faster, semi-automated development style, where you keep an eye on design choices but the code itself gets merged quickly?

It seems to me the project was almost stalled, and now it could suddenly become more relevant: the Activision acquisition could mean trouble for the official Age of Empires. With AI coding we could push the project forward quickly and attract more eyes and contributors.

But it's also understandable if this is your passion project and you want to shape it carefully and personally.
What are your thoughts on the topic?

[TheJJ]

I think games are exceptionally suitable for semi-automated development style, as long as of course quality, maintenance and the architecture are well controlled. That said, I think with our ambitions engine design and keeping it in tact we can greatly benefit from LLM development assistance!

What I want to avoid though is talking to a meat proxy - that is when I as a maintainer have feedback that is then just forwarded to an LLM, i'd rather sooner than later take over the pull request if I want to fix the remaining review comments.
So I would like to expect contributors to be software engineers still, otherwise a pull request is rather a feature wish/bug report, just with some code sample.

Corporate machinations should be rather irrelevant for this project, I'd say - if these things spark more interest in the free software world, I'm happy :)