add guideline on AI contributions#802
Conversation
There was a problem hiding this comment.
Good draft, makes sense to me reading, decision should be with NOMAD management and executive board though, whether we as "The NOMAD Authors" here agree on this policy. I support the here suggested pragmatic approach, specifically the statement in L160 I find good, raising the particular and the non-delegatable awareness on responsibility of humans on AI usage.
The "should" statement in my opinion is too weak to assure provenance of AI usage is collected. Instead, we should rather formulate it strict for now and then professionalize
the wording in the docs throughout to follow e.g. the semantics of "should, must, needs" from a standard e.g. https://datatracker.ietf.org/doc/html/rfc2119
The following of the standard can be a future work.
Making the formulation on the "should" stricter already may help guarding against misunderstandings.
I changed the two "should" statement to "must" now |
|
Push in priority, seek decision with the management, to me an essential PR to merge with the first |
There was a problem hiding this comment.
Provisional approve, provided minor typo gets fix.
Provisional as FAIRmat EB has the final say.
Personally; I think the statement is balanced now,
it is still allowing now the wild west, except, fully AI generated, AI auto generated and merged
Most important though is the statement clarifies responsibilities.
| All AI-generated content must be reviewed, understood, and validated by the contributing author before submission. The contributor is fully responsible for the correctness, quality, and suitability of the code, regardless of how it was produced. Contributions that are evidently fully AI-produced with no human involvement will not be accepted. | ||
|
|
||
| **Code review.** | ||
| AI tools may be used as a supplementary aid during code review, but final review decisions must always be made by a human maintainer. Automated or AI-driven review does not substitute for human judgment on design, correctness, or scientific validity. |
There was a problem hiding this comment.
| AI tools may be used as a supplementary aid during code review, but final review decisions must always be made by a human maintainer. Automated or AI-driven review does not substitute for human judgment on design, correctness, or scientific validity. | |
| AI tools may be used as a supplementary aid during code review, but final review decisions must always be made by a human maintainer. Automated or AI-driven review does not substitute for human judgement on design, correctness, or scientific validity. |
There was a problem hiding this comment.
Most of our documentation uses American English (though we mix it quite regularly, unfortunately), also in the spellcheck. So I think we should leave it at judgment.
There was a problem hiding this comment.
Most of our documentation uses American English (though we mix it quite regularly, unfortunately), also in the spellcheck. So I think we should leave it at
judgment.
Accepted
2 participants
No description provided.