I have never been more burnt out as an open source software maintainer than when people started submitting AI written pull requests. This is hell. I’m in hell right now
@danirabbit Fight fire with fire. I'd put an AI review gate in front of PRs and let it filter out 90% of the slop before a human ever sees it.
@danirabbit Add that to the trivial amount of respect open source maintainers got as it was before.
Its not a recipe for success and positivity.
Its all take take take why why why me me me from some online communities.
Exhausting.
@danirabbit I can only imagine how tiring that would be. Thank you for continuing to maintain the quality of Elementary! It's long been one of my favorite projects.
@danirabbit same. Even just getting them and having to respond with something is just soul sucking.
@danirabbit I had someone admit that they weren't even responding to my comments, it was their AI using their Github account. I had to walk away for awhile.
@danirabbit OSS is in general is exhausting right now because of AI. I have a FOSS project I'm thinking of taking proprietary to secure the source. So it's not trained on. But also, small singles devs rely on donations to keep things going. And people can just git clone and feed it into Codex or something and add the features they want with without developer incentivization to add those features. So you do have to gatekeeper more these days, which is sad.
@danirabbit ditto .. I had to write an 'AI Policy' because of this, and just the amount to review. One PR converts a 30line bash script to a 190line one .. im sure its better in every way, but its a lot to read/understand/test .. and thats just an ancillary script, never mind the code.
@danirabbit I’m so sorry. I can’t appreciate your work enough. 🫶
You’re experiencing the future of productivity! Just let it happen! Let the AI flow into your code base like spice from Arrakis! (The David Lynch one)
@danirabbit Wait until they send you AI slop vulnerability reports.
PRs you can just close, with CVEs not so much.
@danirabbit it drains the life right out if me :(
@janl @danirabbit We're getting a steady stream of LLM-generated or assisted article pitches and now a bunch of pitches for "projects" created using LLMs. I doubt we're seeing as much nonsense as y'all are, but it is so draining.
No. We don't want to publish something written by or assisted by your stochastic parrot.
No. We're not going to write about a software project that you threw together with LLMs and will probably grow bored with when the novelty wears off and/or you run out of money for tokens.
If there's one message LLM users need to hear very loudly, I think, it's this:
"No, you didn't create this. You summoned it. Nobody asked you to, but you did it anyway. You aren't getting a cookie for it. Nobody will be proud of you for doing it. Go away."
@danirabbit And this is why I nearly bit a discredited tech journalist's head off yesterday when he bragged about how much he was having 'vibe coding" some app so he didn't have to learn actual programming.
@danirabbit thanks for all you deligent work and sorry this is happening. I can only imagine how complex and frought reviewing submissions with atypical catastrophic choices hidden in it randomly is. Hope that people respect requests not to submit it or AI collapses sooner rather than later.