Yes, but who is going to verify that every answer is correct?
1 Like
Of course we’re not going to ban wrong answers. As has already been stated, a wrong answer can sometimes highlight a conceptual misunderstanding that other software developers might also run into. If this is due to how the JUCE API has been interpreted, then this is something the JUCE team should look to improve, and it will benefit all users of JUCE. Obviously this feedback cycle needs wrong answers to ultimately be corrected, but I think this happens in most cases when the original poster tries the proposed solution and it doesn’t work.
Again, as has already been raised in this thread, if a response contains some sensibly selected AI output that is fully understood, very well localised on the problem, and tested by the poster, then this has a very similar value to manually authored code. In this case the JUCE team will use their discretion in moderating such content.
The motivation for this change is in the post that started this topic.
Finally, no one in this thread is saying that AI tools are not useful.
11 Likes
In a way it is like protecting the AI against itself: if untested AI examples go into the wild without humans checking the outcome, eventually the AI will eat its own errors, and deteriorate.
A bigger open source project like JUCE is destined to be blowing up in fake untested examples.
I hope that all app devs realise, that dealing with human error is allready enough work: AI hallucinations should be kept out of the equation.
If you can write faster code with it: good for you! but dont let your slave generate code without thoroughly understanding and checking it.
I like to thank all of the JUCE devs for their work: for many of us devs out here, your work on JUCE is close to magic for getting things done, way more important than AI.
3 Likes
AI. 
Secondly, how would you enforce it?
Simple. From now on, everyone has to show their hands in their profile pic.
3 Likes
This is hilarious. Along comes a bot and hammers the point even deeper!
New “users” like the above are becoming a serious problem. I hope the team can find a good way to filter them out!
2 Likes