Basic AI Contribution Policy

[sirosen]

I'm planning on putting in place a basic policy on LLM usage for check-jsonschema. There have only been a few slop PRs so far, but they are persistent, annoying, and sometimes demoralizing -- I doubt they'll go away as an issue.

cc @Julian, if you like, I can do a first iteration there, and if you like it you can refine it into a policy for other repos or the whole org. I'm happy to discuss this as much or as little as you like.

There are various policies in use today, at various levels of strictness. I'm willing to discuss in greater depth, but the basic principles I plan to set down are:

• I can't police what other people do, so I won't.
• Code and prose should be short and precise, not verbose and convoluted.
• Ownership and responsibility are important, regardless of what kinds of tools are used.
• Fully autonomous code submissions are not productive and are banned. (Detecting these is a moving target, but sometimes we know.)
• Don't let a machine think for you, and don't let it talk for you. It's okay if your writing is not "polished".
  • Translation is understood as a special case. Even so, if contributors are comfortable writing in English, at whatever level they are able, that is preferable.