My opinion is you should disclose your process in the repo, but code how you want. You’re sharing your skills and time to us for free.
I’m big on open-source LLMs being the future of LLMs. I’ve been doing a little hobby coding with them, the technology has given me the push to understand things I previously haven’t.
But in a way, it is tantalizing to just “make things work” rather than have it diagnose or fix a specific problem you identify. So, personally, I have slowed the process down to my speed, not to prompt to make another change until I truly understand the changes that were made and checked them.
However, for any kind of creative artistry, writing, dance, music, film, AI has virtually zero good place in the world for me and I know it is doing real harm to artists. For example, music is the artform I’m most knowledgeable in, if AI use could be largely relegated to making Midi files, scoresheets and lyric ideas, I wouldn’t actually have too much of a problem with that. That’s kind of like building parts of code that the artist can synthesize, make melodic adjustments (because AI gen melodies are extremely average) and implement themselves. In contrast, what we have is the wave of samey slop music with the same slop mastering reverb that not only gets on my nerves, but is flooding media platforms with uncreative trash. In code terms it’s like a sea of hackjob programs where you only get the compiled output files drowning out properly developed FOSS.
images