First attempt at a policy for AI/LLM contributions.

[NuclearMonster]

This policy is urgently needed to clarify the project's stance on these tools. We've worked on a few revisions but the commits have been squashed for clarity.

[NuclearMonster]

I'll note that it's my intention to go back after this is merged and clean up CONTRIBUTING.md

The ToC doesn't match the actual contents, and the headings are inconsistent.

[NuclearMonster]

Only fair feedback we've gotten is about specifically calling out human language translation tools (google translate, deepl, etc) to allow their use for certain parts that people rely on since the work is mostly done in English. I think that's fair to specifically allow those for PRs and issues, not sure about code comments

[smallmodel]

How about adding a responsibility part? Something like:

By submitting a contribution, you agree that:
- The work was made by an human.
- No generative AI was used to produce the submitted content.

Would it also be good to structure it like this:

### Policy

Do not submit code, pull request, issues, or other contributions that have been developed with the assistance of generative "AI" [...]

### Principle

This policy exists due to:
- Security vulnerabilities
- Plagiarism, licensing issue
- Respect for unpaid contributors and time
- Environmental impact (enormous impact on limited resources)

### Responsibility

By submitting a contribution, you agree that:
- The work was made by an human.
- No generative AI was used to produce the submitted content.

### Acceptable uses

- Human language translation tools
- Brainstorming, ideas <== ?

This would separate/structure the policy more clearly between the reasoning and the responsibility.

[smallmodel]

Other suggestions for the reason (related to valuable time for unpaid contributors)

• A human can explain tradeoffs they consciously accepted. Generated code can't. If you can't explain why a specific line exists, it shouldn't be proposed at all.
• Humans make mistakes, generated code makes unexamined mistakes.

[NuclearMonster]

This is good feedback, thanks @smallmodel

I've made some changes based on your suggestions.

[NuclearMonster]

Okay, #851 is next.