Hi,
While its not hard to find AI-assisted PRs its hard to set a limit. Even
us, earlier project contributors, use AI to make changes. While it may
not be visible in first run, when I see 10-15 lines of comments for a
2-4 line change its evident that these were left by LLM for whatever
reason. Still a 2-4 line change may fix a bug in a driver.
Since describing a PR is not an issue, cause agent do it pretty well,
with "artificial" restriction to copy same text to mailing we risk, that
eventually same slope will end up there. Sure, its nice to increase a
traffic in mailing list, but maybe with more valuable conversations,
than AI generated content.
To recap - its hard to define a single standard, but having a warning on
contributor guidelines about heavy use of AI, in nowadays conditions, is
a must.
Best regards,
Łukasz
On 8/29/26 13:57, Christofer Dutz wrote:
Hi all,
As you might have seen we’re getting more and more random PRs generated by
assuringly AI generated drive-by contributors only wanting to gain „reputation“
by showing off their „hard work“.
I would propose to update our contribution guidelines, that people submitting
PRs also are required to show up on the dev list and introduce themselves and
tell us about who they are and why they want to contribute.
If someone doesn’t do that I would like to start simply pulling in legit
changes without „awarding“ them a countable contribution.
If I hear no objections, I’ll update our web content accordingly when doing the
updates prior to the release next week.
Chris
