AI/LLM contribution policy

[mahrud]

I've noticed in the past few months several open source projects have had debates about dealing with LLM-assisted contributions, with widely diverging decisions; e.g. QEMU bans it outright, out of concern for future legal issues with copyright, while others like Fedora and LLVM regulate it.

I think it's fair to say that there's already some LLM-contributed code merged in M2, but presumably it won't be long before we have to review package pull requests with LLM written documentation. I think it would be good to establish some ground rules ahead of time and advertise it in our pull request template.

[d-torrance]

Great idea! I'd be in favor of a policy modeled after Fedora/LLVM, as I find AI assistance quite helpful. I haven't used it to write any Macaulay2 code yet, but it's great for things like:

• brainstorming ideas for package design (e.g., the JSONRPC package took shape by bouncing ideas off of Chat GPT)
• writing repetitive documentation nodes (e.g., many of the Python documentation nodes in Bump Python package to v1.0 #4025 were initially generated by Copilot using copilot-mode in Emacs, and then usually modified a bit by me afterwards)