New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Debian votes to let contributors code with AI
Just saw this
Thanked by 1forest

Comments
The end is nigh.
This:
“The use of a generative AI tool does not diminish the contributor's responsibility for the work they submit. Contributors are expected to understand, review, test, and, where appropriate, modify AI-assisted output before incorporating it into Debian.”
Essentially it says "no matter how you create code, in final you are responsible for it". So devs ass can burn if shit gets pushed. Acceptable.
You can't simply deny LLM. It is here and it will stay for decades.
I don't mind it. They allow responsible use of AI tools, which gives them enough leeway to prohibit slop.
There are some good comments on LWN: https://lwn.net/Articles/1091231/
Well the debian guys just confirmed my sig. I'd blacklist them anyway if I hadn't already for the poettering shitfuckery.
They can prohibit as much as they want but..
can they enforce it?
So they do like Pontius Pilate and say basically this: "We wash our hands of it"
shit will get pushed, make no mistake.
Time to pick a new distro.
You must be playing the long game my friend! You're in the majority now, but in a few years your views might be exotic!
In the end, they can't tell if the code X dev wrote in vscode was enhanced by AI or not (or even 100% AI).
Debian is stability standard. “Another distro” doesn’t cut it. Ubuntu smells, arch missing in marketing and centos no more. Bsd - fck that, to complicated.
I’am with debian since 3.0, pounded here hard. It always endured. Servers died, but os spinned to the last second.
i always use debian for production-grade servers, and what are they doing here
I do not care about what the majority "thinks" or what's considered exotic etc., at all.
AI bad, hooman good (2026)
Computer bad, hooman good(1980)
Tractors bad, horses good(1950)
I am not sure why people hate it when most software guys use AI for most of their coding sessions. Things have changed rapidly(a lot) in last 4-5 months and skipping it would be a big mistake 😄
AI assistance and vibecoding are completely different levels.
This is good, and if you think otherwise you're living under a rock and will be left behind.
I think AlmaLinux cuts it. In fact I think it may cut it even better than Debian does. A lot better!!
AlmaLinux may not support low end hardware as well as Debian does because it is derived form RHEL and only x86_64-v3 is supported there, many LET providers still use non v3 hardware.
However, AlmaLinux does offer x86_64-v2 versions designed for older hardware and that should work on most LET providers. The v2 versions allow continued security updates for legacy systems but may have limitations with third-party packages which may have been built for v3.
I'm experimenting with AlmaLinux, I think it can play nicely with most LET providers.
I like Debian and lately AlmaLinux tick all the boxes
For me it's Alpine, one main reason being that I think that lots of the linux distro bloat shouldn't be on a node anyway. Debian highly likely was good once but they've demonstrated more than clearly how they tick (woke and weird) when they fell for the systemd cancer and frankly, I'm not even surprised that now they also joined the ai cult.
I've been relying on a mix of Debian and Alpine but from now on I'll be jailing Debian.
So for me it will be a mix of AlmaLinux + Alpine + Jailed Debian
The end of an era I guess, however getting rid of Debian will take a very long while.
They didn't join the AI cult. The results of the votes and discussions show that they're still extremely skeptical about AI. Their decision was essentially just that contributions are to be judged on their own merit, meaning vibe-coded slop will be rejected. If someone doesn't deeply understand the patch they're proposing, it'll be rejected.
This.
Plus, I'd want any distro to be thoroughly pentested with AI assistance to find vulnerabilities before releases as well.
I'm not so sure why we're having AI find critical vulns left undiscovered for a decade in code, but many will argue you shouldn't use the same AI to build or secure that code.
Textile machines bad, child labor good (1810)
Yeah whatever, AI is useful, just look through the results and it's all fine.
A skilled person using AI as a tool to assist = good
Unskilled person using AI to vibe everything = you already see/know the results
Different tasks and very different mind sets! One ideal for computing, the other one not at all.
Nope. My line is not "I like the old way, I hate the new way"!
All others "were good" because they were helping/supporting while us (mostly *g) being in control - except for ai which is mimicking us or, worse, replacing or even sooner or later killing us.
Gladly though it needs so much valuable resources that "agi" will not be achieved, and today's ai bubble will burst (but SLMs - as opposed to LLMs - might continue to "evolve" and survive).
Btw, how is quantum computing doing? *g
As if ... plus, yeah sure, trust something that's known to hallucinate and to lie.
Ubuntu smells? rude. Ubuntu on all my machines.
Improving at an unsettling pace to the point that I'm trying to cover my ears and convince myself that Gutmann has a point.
Simple as someone who has claimed over 10 bounty's from various programs, LLM's are used because humans rely to much on luck, or the obvious. When bug / exploit hunting you rely on your starting point to take you to the end. Humans aren't great at consuming 100's of thousands of lines to find the 1 particular issue. We are creatures of habbit we go to what we feel is obvious, or leads to something over and over.
An LLM will score the entire codebase, check 100's of packages and 100's of thousands of lines. Honestly i'm surprised there has not been more CSVE's and I suspect this is mainly due to the limitations of truly good hunting models being limited from public use currently.
*g
I don't know what *g means.
But now that Kimi K3 is out...
its fun to rage about ai but the hammer doesn't make the carpenter, stop being silly.