When working through the backlog of PRs recently, it's pretty clear that many of the open PRs are LLM-assisted drive-by contributions that are either from folks new to Python packaging or even completely automated without a human in the loop. This isn't very productive. Hence I'd like to propose a few changes:

• Remove the good-first-issue label. Having this label was good practice 5-10 years ago, it's considered bad practice by a lot of OSS maintainers today. It attracts a lot of low-quality PRs and almost never leads to contributor retention.
• Add a PR template which asks for two things that help maintainers assess PRs:
  • A "What motivated this PR?" question. This tends to be really helpful in my experience: an answer like "I encountered this real-world problem and looked for an answer in these docs but didn't find it" tends to be a positive signal and a "I saw this open issue and thought I'd have a go at it" is a negative signal (especially when coming from a new contributor).
  • An "AI Usage Disclosure" section. Disclosing of AI tool usage, and for what and which tool/model, is good practice.
• Update the contributing guidelines with expectations for AI tool usage. It's fine for code or review I'd think (and quite a few PRs using AI tools already went in), but not for writing a PR description or replying to reviewers - as already pointed out by other maintainers on a number of PRs, the expectation is that we're interacting with humans who write in their own words, showing they understand what they're submitting and saying.
  • We don't need to go overboard here, we can just state these basic principles, or copy a relevant policy from this long list of examples - no need to reinvent this particular wheel.