Am 01.06.26 um 19:42 schrieb Dan Eble:
On 2026-06-01 03:49, Lukas-Fabian Moser wrote:
Of course, since the number given will (according to your proposal)
influence how the MR is dealt with, we depend on getting an honest
answer to that question. I don't want to seem paranoid, but maybe it
would be wise to add - somewhere in the CG - a statement along the
lines of: Commits with non-disclosed AI-generated code get refused
(or may get reverted later).
I'd perceive that notice as unwelcoming. If the answer is missing we
could just draw attention to it gently. I'm sure someone will
eventually accidentally delete the question while editing their
summary and need just a nudge about it.
If we ever find something that shouldn't have been merged, we should
just revert it. We don't need to give notice.
I wasn't concerned so much about missing answers than about stricter
endorsement demands for AI-generated MRs possibly being an incentive to
not disclose the AI involvement truthfully.
But your point is fair, of course: We should neither be unwelcoming, nor
should we assume the worst in people.
In the discussions of the current MR I noticed the term
"AI-assisted". Maybe it would we a good idea to distinguish various
kinds and degrees of AI assistance: IIUC, not every way of using an
LLM during development leads to longer, coherent blocks of AI
generated code.
Good point. I think it would be best if the question in the MR
template didn't demand a lot of thought to answer or interpret. I
wouldn't want people splitting hairs or making accusations like, "but
you said 50%, this is closer to 75%."
I suggested "AI-generated" because "AI-assisted" sounds broader than I
care about. I don't care if someone has used AI to analyze code and
explore solutions. I care about what I am reviewing. If it's
generated content that hasn't been under a critical eye yet, that's
important.
Agreed. These distinctions and explanations would probably be a good
addition to the CG eventually, though.
Lukas
