Hello, I am concern about security with all this rejection of AI in the foss world, I am a foss user and will remain and defending its use. I don’t like all this vibecoding wave, but using AI to defend us be it finding security bugs using AI or correcting them appointed by others that use AI to discover and signal CVE’s and bugs. Is f-droid against this use of AI?
nice to know debian aren’t ai doomers, gcc a bit of a letdown, sad that they were forced to participate in this buzz at all
I think there’s a strong anti-AI stance that’s making it hard to agree on governance/manifesto of AI use.
At work, I’m encouraging the following with our developers:
- Own your commit by verifying correctness
- Prove your work is safe
- Prioritise quality over quantity
- Document intent - write clear and meaningful commit messages, and document architecture
- Be aware of licensing boundaries
EDIT: Found this Human OSS.
Okay my point to these asks: people who cannot or do not write code themselves and use AI, how can you expect them to bother and write anything else? Documentation, security, they will or are doing all with AI. Trust them, NO way.
This is right here why i do feel like Stricter rules need to be enforced on AI usage. I don’t mind a developer using it to help with certain things though i’d rather them not use it at all. But i get it, it can genuinely help with easing certain tasks and operations.
However i genuinely do not like when people just use it like some sort of magic wand with little to no understanding of how it works or what the code it is writing actually does.
i did try an experiment one time as an example. I maintain my own uBO filterlist for hardening and personal use. (guess you could say it’s propietary cause there’s not an actually pubicly available source.) So i decided to see how AI could stack up against it. I asked AI to make me a filterlist. Wasn’t planning on using it. Just wanted to see.
And overall it did okay it was a decent filterlist… Just one problem. There were a good amount of rules in there that were just plain invalid or worse were actually harmful for one reason or another. Of course i could fix that since i am very familiar with how uBO’s syntax and filter syntaxes in general work but imagine if i wasn’t fimilar with that and i decided to freely and openly open a repository for that for all to use. Would have been bad.
But even still after that i don’t use AI in my filterlists. The work is 100% done by me and for one main reason. Using AI… Just doesn’t feel rewarding at all to me. I like maintaining the list myself cause not only is it fun and rewarding in the sense that i actually did something very productive spending effort to maintain a list that makes the web less harmful and less annoying. Additionally doing it completely myself means i actually learn from my experience and does ensure i can produce a more quality list.
Sorry for the long.. Whatever you want to call it i just wrote i just wanted to say how i feel from experience.