[QGIS-Developer] Floating an idea: ban AI based contributed from non-core developers?

Sun Apr 12 09:51:05 PDT 2026

Hi Harrissou and all,

Bien cordialement,
Régis Haubourg

On 12/04/2026 13:25, DelazJ via QGIS-Developer wrote:
> Hi again,
>
> Nyall, I'm sorry to see that my message touched a nerve. Sorry. I 
> reread myself and am (un)fortunately unable to find what part of my 
> message triggers that:
> - the technical aspect of welcoming was about making sure we provide 
> correct and up-to-date materials e.g., it's regular that the 
> install.md file is outdated. It was in no way a call to open doors for 
> AI tools.
> - on the interaction part: I didn't mean we welcomed warmly any 
> AI-based (attempt of) contribution either. My point was to avoid being 
> unwelcoming with real potential contributors just because they used 
> AI, something the repo guidelines allow them to do. A neutral label as 
> Even updated it to seems a good start, and we can imagine applying the 
> "slop!" one to the PRs that reveal to be of low-quality, allowing us 
> to quickly close and ban the authors after a number of recurrence.

The topic of open source future in AI's world worries a lot of us. QGIS 
is both a professional tool, but also a passion for many of us and I'm 
happy that emotions and feelings about AI mis-uses impacts generate 
debates. Thanks all for keeping constructive and peaceful like you do. 
That is why I love this community.

>
> Anyway, sorry for any misinterpretation my message would have led to.
>
>
> Greg, there are two different things: people submitting low-quality AI 
> generated code (and these are problems we need to solve quickly), and 
> people understanding what they do and using AI to help them polish the 
> work. I personally probably don't have skills to judge who belongs to 
> which group but this could be a first step of review; I'll leave that 
> to others. My key point as explained above is that, as long as we 
> allow (for good or bad reasons?) AI assisted PR, the second group does 
> not violate our rules, so we have to be welcoming with them. That's all.

One key point here. It is really **hard** to decrypt the quality of LLM 
generated code, because it is trained to be plausible.

  Knowing this, a friendly contributor, willing to getting involved in 
the long term, and who will use AI as a helper for its first 
contributions, should know that it will put a very high charge on 
reviewers. So this is like a poisoned gift.

My personal view is that contributors should take care of showing they 
are human, by describing their real motivations, plans for the future. 
Any purely technical PR, launched from an anonymous account, out of the 
blue, was already a not-so-good practice. In AI era, this is a very bad 
practice and I agree this is already "AI-slop". A presentation message 
on the developer's list and some context in the PR is the bare minimum 
to me.

  We're collaborating with each other. We appreciate meeting in 
contributor's events from time to time. We also like the project and 
like to take pleasure in doing good work, event when paid for not-so-fun 
tasks.

This idea of being part of a community is our real strength. Let's make 
it clear that using AI is one tool only, it does not replace human 
relations.

>
> Kind regards,
> Harrissou
>
>
> Le 11 avril 2026 21:08:36 GMT+02:00, Greg Troxel <gdt at lexort.com> a 
> écrit :
>
>     Nyall Dawson <nyall.dawson at gmail.com> writes:
>
>         Advance warning: Emotional reply. Why, as a society, can we
>         recognise that it's generally a BAD THING that AI has
>         destroyed the livelihoods of graphic artists and musicians,
>         but are BLIND to the impact it is having on professional
>         software developers? Why MUST we be welcoming and warm to the
>         people who are vandalising OUR form of art, and making it
>         impossible for us to continue in the industry that we've
>         devoted our lives to? I'm sorry, but I have absolutely NO
>         sympathy for a new ai-slop contributor to open source. Just
>         like I have no sympathy toward someone flooding spotify with
>         AI written music, making it impossible for real artists to
>         make a living. Screw that. It's already thankless enough to be
>         an open source maintainer. Now it's even worse. Look after the
>         people who have devoted YEARS of their lives to open source,
>         or you'll lose those. /me out Nyall 
>
>     Before I saw your reply, I was going to comment that the "it's
>     unkind to label low-quality LLM-generated content as AI slop"
>     statement, while coming from a place of wanting to be welcoing, is
>     missing the point that submitting low-qualitty LLM-generated
>     content is an offense against the community. We would not be
>     welcoming to people posting intentionally offensive comments
>     (e.g., gratuitous off-topic racial epithets) to issues, and we
>     wouldn't be talking about how it was unwelcoming to just delete
>     them and ban the submitter. This is less different than people
>     that think AI is ok think. It is possible that some people
>     submitting AI slop don't undersetand that it is AI slop, and don't
>     understand the harms that LLM-generated content does to projects.
>     Once there's a clear place to point to -- which explains that the
>     submitter should not again submit other LLM-generated content --
>     it might be better to just close with a pointer, rather than label
>     AI and leave open.
>
>
> _______________________________________________
> QGIS-Developer mailing list
> QGIS-Developer at lists.osgeo.org
> List info:https://lists.osgeo.org/mailman/listinfo/qgis-developer
> Unsubscribe:https://lists.osgeo.org/mailman/listinfo/qgis-developer
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/qgis-developer/attachments/20260412/9d6c1415/attachment-0001.htm>