Hi all, I was just asking myself metaphysical questions about the contributions to QGIS …
The PR template now includes an “AI use disclaimer”, which I personnaly think is a nice and honnest thing to repport … but some recent contributions that fair-played were “harshly” tagged as AI slope before any real review (Fix/63672 fix plugins version issues by T4mmi · Pull Request #65630 · qgis/QGIS · GitHub, Add geom_from_geojson function to field calculator by marcu · Pull Request #65573 · qgis/QGIS · GitHub ). I totally understand and share a lot of concerns about the “fear” of AI slope PR, but maybe it lacks nuances in the approach ?
I think there is a (rather huge") margin between a PR made by a human who used a LLM prompt to find its way into a 1000+ lines file or just the gigantic qgis API, and then wrote some code (even AI-assisted) to fix a bug or develop a feature … and a fully-autonomous claw agent PR thats “fixes” who-knows what.
Again, I’m one of the skeptics about vibe-coding, yet I see many developpers relying on it more and more, and past the “old fool” reaction (which I have too) I think it’s “just a tool” that at least deserve some credit for many tasks (like find an entrypoint for a fix, feature … or just diggest some code mecanisms) …
and at least with the actual behavior, I fear that people will just hide their use of AI rather than “not use it as it’s evil” … (I know i won’t mention it again … might be a stretch but I think any “intelligent” thing, artificial or not, would do the same after a try or two)
Any thoughts ?
[edit] the label was changed to “AI assisted”, wich is a bit smoother, yet the point remains …