Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Debian votes to allow "responsible use of generative AI"

lwn.net

Post reply on HN
Print view View original

351–360 of 484 posts

Re: Debian votes to allow "responsible use of generative AI"

#351

Post by inigyou »

b112 wrote: post #248
lrvick wrote: post #200

Earlier quoted context omitted.

I use agents to build exactly what I want like I am pair programming with a jr engineer, and review every line of final output before I make a PR. I would have zero tolerance if anyone that did anything less in my orgs. Owning and understanding the code you submit is the one job that matters most now.

I feel that in a few years, or even less time, we'll see a whole bunch of companies implode, whist others remain fine.

We should have already seen it. Software was slop quality even before AI. But we didn't see it.

Re: Debian votes to allow "responsible use of generative AI"

#352

Post by JohnHaugeland »

KennyBlanken wrote: post #218
Aurornis wrote: post #164

Earlier quoted context omitted.

Yes. There are pockets of hardline anti-LLM policies for various reasons. SourceHut bans LLM generated code from their platform https://news.ycombinator.com/item?id=49461724 The Asahi Linux project bans LLMs. Unfortunate as LLM reverse engineering and driver writing would bring Linux to more Mac models much faster. The Zig project rejects LLM based contributions. Some of these policies were introduced when LLMs were…

> Some groups don’t care about the quality and object on moral or copyright grounds. Precisely why Debian Project's decision is both baffling and infuriating. I hope Debian forks over this. They have severely lost their way.

if it does, that fork is dead by the end of the year

i don’t think you understand just how niche you already are

Re: Debian votes to allow "responsible use of generative AI"

#353

Post by kccqzy »

stillpointlab wrote: post #310
GZGavinZhao wrote: post #110

Related: I find this self-assessed AI level for contributions [1] to be extremely useful, both professionally and personally, to communicate the level of AI assistance so that the person on the receiving end can evaluate how much time & effort they want to spend understanding my code. [1]: https://www.visidata.org/blog/2026/ai/#self-assessed-ai-leve...

I think 10 levels here is too much and hints that someone anchored on 10 as a round number and then forced the levels to fit after the fact. I don't really see much value beyond the 3 levels of AI disclosure that have been floating around: - fully human - ai assisted - ai generated

There’s a big difference between AI generated code that the human understands completely, the human understands mostly, or the human understands not at all.

Re: Debian votes to allow "responsible use of generative AI"

#354

Post by sarchertech »

Loughla wrote: post #273
bfgeek wrote: post #263

Earlier quoted context omitted.

The issue that open source projects are facing at the moment is that it takes significantly less effort to submit a patch for review. A lot of developers who are submitting these AI patches don't necessarily understand the patch, so the onus is on the reviewer/code-owner. The reviewers are getting swamped (some reviewers are receiving 100s or patches per month). If feedback is provided at lot of the time the patch au…

The question becomes, does it take more time to create or review in the Grand scheme of the software life cycle? Because if it's still a time saver, even with the increased review load, then it's a win, correct? I'm not a coder so I have zero idea. Thoughts?

Well historically the consensus was that it was harder to read code than to write it, but that leads to uncomfortable conclusions these days, so that bit of common wisdom has mostly been ejected.

> I'm not a coder so I have zero idea. Thoughts?

I’m not trying to be mean, but this comment is basically “I have no experience with this topic, but it can just be boiled down to this one simple question right?”

Re: Debian votes to allow "responsible use of generative AI"

#355

Post by inigyou »

Zambyte wrote: post #337
fortzi wrote: post #294

Earlier quoted context omitted.

Claude won’t tell you why it did something. Instead, it will generate a probable explanation. The two are very different

I'm not sure how context is managed between reasoning traces and actual output in Claude / claude code, but if the reasoning trace is in the context of the question for "why did you do that", it can actually answer why it did that.

Reasoning traces are also probable explanations.

Re: Debian votes to allow "responsible use of generative AI"

#356

Post by fortzi »

Zambyte wrote: post #337
fortzi wrote: post #294

Earlier quoted context omitted.

Claude won’t tell you why it did something. Instead, it will generate a probable explanation. The two are very different

I'm not sure how context is managed between reasoning traces and actual output in Claude / claude code, but if the reasoning trace is in the context of the question for "why did you do that", it can actually answer why it did that.

I may be mistaking, but I doubt it digs through thinking tokens of previous runs, not to mention previous sessions

Re: Debian votes to allow "responsible use of generative AI"

#357

Post by inigyou »

createful wrote: post #129
bigstrat2003 wrote: post #57

Earlier quoted context omitted.

Except LLMs actually are making programmers dumber. I personally know people, people whom I know for a fact were good programmers before, who have now completely stopped using their brains. When you ask them why they did something in the code, they say "I don't know, Claude did that and I didn't really question it". There was no such analogous phenomenon with IDEs.

You could also say the same with desktops. Desktops abstracted CLI commands like mv or ls. People who are taught on desktops will get scared when they see a terminal. Yet programmers will still know commands. Abstractions can make things easier but it doesn't require losing skills.

Desktops don't abstract CLI commands. Both are different interfaces on the same thing underlying. Both coreutils and Windows Explorer are "abstractions" (as you call them, but they're actually interfaces) on filesystems.

Re: Debian votes to allow "responsible use of generative AI"

#358

Post by dooglius »

eterm wrote: post #315
dooglius wrote: post #300

The voting seems to have been pretty much linear to how pro-LLM they were. So it's interesting that all of the proposals were essentially anti-LLM and the chosen one was the mostly neutral, only slightly anti-LLM one. The absence of any pro-LLM proposal indicated that proposal writers are not in sync with the voters

What would "pro-LLM" look like to you?

Ex: Security-critical code contributions should be scrutinized via state of the art tooling, including but not limited to fuzzers, linters, and adversarial LLM review. For non-security-critical code, use of LLMs is encouraged but not required. The bar for code quality should be raised as the skills of coding agents increase; a developer's choice not to use LLMs is no excuse for miskates that an LLM would catch.

Re: Debian votes to allow "responsible use of generative AI"

#359

Post by Topfi »

rapind wrote: post #277
Topfi wrote: post #181

Earlier quoted context omitted.

For me, it’s not a question of whether I like it, nor where we might move in the future, but purely what is proven currently. In my experience, as in my evals, any model, including notably Fable 5 and GPT-5.6 Sol can and indeed do produce code that does not consider many edge cases, can be hard to maintain (especially Sol though it’s output has become markedly better with recent changes to Codex) and sometimes introd…

It's not about the prompt, it's about the guardrails. Static types, compiled, massive test suite, etc. That what I mean by extreme level of test coverage like SQLite. You'd actually need to be an expert in C and SQLite and the SQLite test suite just to sneak a bug into it. Testing and other verifications are going to be an exciting area for the foreseeable future. You can still write non-performant code even with tes…

Am about as far away from a C expert as one can be (and probably want to keep it that way for the time being), so happy to take your word on that front. Still feel, it'd be easy and in their interest for one of the major labs to provide a verifiable showcase/research paper on that front. Zig-to-Rust in Bun is the closest and that is very impressive, but I feel it doesn't necessarily map to new features/changes beyond porting. The compiler, browser, etc. all equally do not fill me with that much confidence, but perhaps in SQLite modifications could be hard for a model to mess up.

What I will say is that I feel valid in saying that the developer behind gas town (a shudder runs down my spine even having to think about that) is not working in a test suite anywhere comparable to SQLite, though in fairness, you never claimed that.

Again, you are likely not wrong that with an environment like SQLite, bugs may be hard to sneak in, but then again, isn't retaining that so important that it is a strong argument for developers auditing LLM output even more strictly, lest we loose the resilience we currently have?

Re: Debian votes to allow "responsible use of generative AI"

#360

Post by seizethecheese »

Could it be that coders are overly focused on the method for generating code over other forms of slop?

For example, I’ve seen senior devs outsource their thinking entirely to Claude, then carefully (slowly) generate code that’s not slop on a per-line basis but is slop on an architectural level.

Post reply on HN