You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Proactive guidance for AI-assisted contributions
#68
This is a proactive discussion, not a response to any existing PR. It should not be read as suggesting that any prior or current contribution is low-effort. ReEDS has not yet received many external contributions, which makes this a good time to set expectations before ambiguity creates friction.
Intent
The intent is not to ban AI tools.
The intent is to make sure contributions remain:
reviewable
accountable
understandable by the contributor
respectful of maintainer and reviewer time
Proposed principles
1. Human accountability
The contributor is responsible for everything submitted, including code, data changes, documentation, issue text, PR descriptions, and review comments.
2. Human understanding
Contributors should understand their changes well enough to explain the motivation, implementation, expected model impact, and testing approach.
3. Self-review before maintainer review
AI-generated or AI-assisted content should be carefully reviewed, edited, and tested by the contributor before asking maintainers to review it.
4. Respect for maintainer time
Contributions should be scoped, documented, and tested well enough that review is a reasonable use of maintainer time.
5. Transparency
If an LLM-based coding assistant or chatbot was used in a meaningful way, contributors should briefly describe how it was used in the PR description.
6. No autonomous submissions
Bots or agents should not open PRs, issues, comments, or review feedback without human review and approval.
What this should and should not do
This should
This should not
Set expectations before we receive more external contributions
Accuse existing contributors or PRs of being low-effort
Permit useful AI assistance with human review
Ban all AI-assisted work
Make review expectations clearer
Add heavy process for small contributions
Protect maintainer time
Discourage thoughtful new contributors
Ask for transparency when AI meaningfully shaped a PR
Require disclosure for trivial grammar or formatting help unless we decide otherwise
Possible wording for CONTRIBUTING.md
Draft text
## AI-assisted contributions
Contributors may use AI or LLM-based tools when preparing contributions to ReEDS, but a human contributor remains responsible for everything submitted.
Before opening an issue, pull request, proposal, or review comment, contributors should review and understand all AI-assisted content. Contributors should be able to explain the motivation, implementation, expected impact, and testing approach for their changes.
Please do not submit unreviewed or lightly reviewed AI-generated content. Contributions that are inexpensive to generate but costly for maintainers to review may be closed or sent back for more preparation.
If an AI or LLM-based tool was used in a meaningful way while preparing a pull request, briefly describe how it was used in the PR description.
Automated agents or bots should not submit issues, pull requests, comments, or reviews without human review and approval.
Possible places to update
CONTRIBUTING.md: best place for the main policy because it is visible to new contributors.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Proactive guidance for AI-assisted contributions
Note
This is a proactive discussion, not a response to any existing PR. It should not be read as suggesting that any prior or current contribution is low-effort. ReEDS has not yet received many external contributions, which makes this a good time to set expectations before ambiguity creates friction.
Intent
The intent is not to ban AI tools.
The intent is to make sure contributions remain:
Proposed principles
1. Human accountability
The contributor is responsible for everything submitted, including code, data changes, documentation, issue text, PR descriptions, and review comments.
2. Human understanding
Contributors should understand their changes well enough to explain the motivation, implementation, expected model impact, and testing approach.
3. Self-review before maintainer review
AI-generated or AI-assisted content should be carefully reviewed, edited, and tested by the contributor before asking maintainers to review it.
4. Respect for maintainer time
Contributions should be scoped, documented, and tested well enough that review is a reasonable use of maintainer time.
5. Transparency
If an LLM-based coding assistant or chatbot was used in a meaningful way, contributors should briefly describe how it was used in the PR description.
6. No autonomous submissions
Bots or agents should not open PRs, issues, comments, or review feedback without human review and approval.
What this should and should not do
Possible wording for CONTRIBUTING.md
Draft text
## AI-assisted contributions Contributors may use AI or LLM-based tools when preparing contributions to ReEDS, but a human contributor remains responsible for everything submitted. Before opening an issue, pull request, proposal, or review comment, contributors should review and understand all AI-assisted content. Contributors should be able to explain the motivation, implementation, expected impact, and testing approach for their changes. Please do not submit unreviewed or lightly reviewed AI-generated content. Contributions that are inexpensive to generate but costly for maintainers to review may be closed or sent back for more preparation. If an AI or LLM-based tool was used in a meaningful way while preparing a pull request, briefly describe how it was used in the PR description. Automated agents or bots should not submit issues, pull requests, comments, or reviews without human review and approval.Possible places to update
Open questions
References
All reactions