Please do not misunderstand my criticism below. I regard the current craziness about "AI" to be a huge issue in many ways, ethical, societal, cost-effective etc. And I see the huge and unfair force that its abuse creates on maintainers and developers. I desire it all had never happened but current we are.
I desire to emphasize several technical benefits though, which this guideline would disallow, at smallest as I comprehend it. I am deliberately ignoring everything alternatively here.
I could additionally entirely comprehend if it was additional of a "political" move to ban item considered unethical. But is that the intention? Is it to decrease force on developers? Or to create a powerful declaration against "AI"? In any case, I think it could additional apparent and likely be stronger.
As it is written now, I would comprehend the guideline to not authorize usage of LLM for review, auditing and safety purposes either. The words assisted and used are extremely broad and I am fearful that they power throw out the baby alongside the bathwater (if the baby is AI use that yields things that are functionally and logically fit for integration in QGIS, that are human-reviewed and that do not create additional after-merge-maintenance-hassle than those 100% originated from individual brains).
As person who sometimes feels brave adequate to create small steps into the C++ earth of QGIS core, I have established LLMs to be helpful for discovering additional, non-obvious environment in the code. They are frequently wrong, but it is my powerful condemnation that the results can be used by humans to acquire additional understanding of code and really assistance creating improved contributions.
I additionally sometimes discover LLMs helpful to get a simulation of a reviewer. Throw your code at it for review, get several listicle back, discard several of it refuse or unrelated, appearance at the rest, perchance discover an genuine issue, fix it, afterward go to your co-workers alongside a slightly small bad PR. A "Clippy" in your IDE, that gives several - sometimes helpful - feedback or criticism on your development, can additionally assistance decrease errors and enhance robustness. Like a web hunt volition output false data alongside true information, a LLM volition do the same. It is up to the individual to critically verify any they output.
Such uses would be disallowed by my understanding of this policy.
