I had to use a vibe coded library for a personal project as the alternatives just didn’t work, and I don’t care about the bugs as it’s just for me to run the app one time.
However I did find some missing features I required that were easy to add, so I forked the project and added them. Then made a pull request.
But that made me think, would fellow anti ai people actually contribute back? Even if it was originally made with ai? Or would you go to the extreme and rewrite it yourself?
Update: after some reflection, I realised that it was hypocritical of me, as I got a rule of not promoting AI made content. I will not remove the pr, as they can just reopen it, but won’t do it again
No. Engineers today are over-reliant on libraries, which is enough of a risk without bringing in AI vulnerabilities and oversights. If there isn’t something that is already widely adopted and well maintained, I would just write it myself.
Yeah, how dare c++ use boost or the stl, or Python devs use the requests or sqlalchrmy library. 😐
I wouldn’t contribute back, and would switch to (or write) an alternative ASAP.
There are cases where I have to use AI-tainted dependencies (though, none of the AI-tainted dependencies I currently use across my projects are fully vibe coded, they’re merely tainted), but if I have to patch one? That’s gonna be a fork or rewrite, and there’s no chance in hell I’m contributing back.
I mean, if you already fixed bugs or improved features, then you might as well raise a PR.
Vibe coding is an indication of low quality and longevity, with a higher likelihood of bugs, flaws, and vulnerabilities. Trust takes time to build, and I’m sure some vibe projects will turn out okay, but they will always require extra scrutiny as any idiot can ask an LLM to code them anything.
Depending on your definition of vibe coding, no probably not.
Setting aside for a moment my feelings on the current state of “AI” and AI coding, I don’t see that there would be a point. The next vibe refactor is likely to overwrite any changes I write so it’s unlikely to be useful to either the project or me.
They probably can’t.
Arguably I’m creating a vibe coded repo now.
We have a fairly complex continuous integration environment but need to move to GitHub, because we like following fads and don’t realize that fad is over. But GitHub CI is really immature, difficult to use, and just can’t do anything beyond the basics. I’m sure it’s great for individual projects and startups but it just doesn’t scale for enterprise use. Everyone gets excited at using yaml to describe workflow, but then you’re kind of stuck with shell scripts for actual implementation and the easy stuff runs out after the first hour. I am very much not impressed.
Engineers are excited over all the “candy” they see opensource projects use, and don’t understand most of that will be blocked because we’re an effing security company and you just can’t use random code
So I’m trying to create a shared workflow for like 5,000 repos and it’s miserable. After a couple weeks I have a couple thousand lines of generated shell script that’s already beyond my ability to maintain to maintain by hand. So I’m well past “I told you so”, and my only route forward is yet more generated shell script. It’s too late to try to contribute without ai
