Thinkin’ about LLMs? That’s a paddlin’.
Then there is bun where the rules are probably written by an LLM and say “all non LLM uses are forbidden”.
Context: The did a complete rust rewrite with LLMs that was “just an experiment” and is now probably being merged.
Good luck enforcing no LLM spell check
Reading between the lines they wrote this because of seeing “I hear you, and you are right to be suspicious — but I only use AI for spell checking, nothing more” one time too many.
Just skimming this thread (I don’t even know what Zig is) and this automatically flagged something in my mind. Ugh. Make it fucking end already
No people who have ever used, or heard of LLMs.
or heard of LLMs
This instantly disqualifies anyone who has ever read these rules.
I find the “no use of AI for finding bugs” weird, so long as you’re also confirming it yourself.
so long as you’re also confirming it yourself.
The issue is that 90% of them don’t.
They throw your code at an LLM, the LLM finds ‘bugs’ that may or may not actually exist (and may or may not be duplicates of other reports), the LLM writes up bug reports for them (always marked as urgent, top priority, security flaw), and those bug reports are posted without the user reading or understanding them or even being capable of understanding them.
For every bug reporter who’s using an LLM responsibly, there are 10 more blindly throwing bullshit like this at you.
The book by Asimov mentioned there is really good
They hiring?
If the Zig community carves out a territory of principled engineering like this, I may adopt it as my primary language and make a career out of it. Finally an island of sanity in a sea of slop.
It’s beautiful
Zig is really moving on this one.
It should be noted that e.g. DeepL which is a very good AI translation service isn’t an LLM but rather falls into the category of “Neural machine translation”. So this would still be fine
LLM stands for “large language model” in other words, it is a big neural machine. Saying they don’t use LLMs is like saying the ocean isn’t blue, it’s azure.
An LLM is a type of artificial neural network, but not every ANN is an LLM. So its a lot more like saying that not every red thing is a firetruck.
That is an extremely simple concept that even toddlers can understand.
LLMs work by increasing scale, number of separate neural networks, to increase accuracy when improvement from training hits a wall. Which is very problematic because it means power consumption becomes exponential. I think most people won’t have a problem with neural networks, but certainly do have a problem with LLMs.
Machine learning, neutral networks, AI, in general it’s very useful when trained at a specific task. LLMs are most certainly where things went wrong.
DeepL uses plenty of LLMs internally and recently laid of around 1000 employees to “shift to AI”.
To be fair, LLMs do really good translations, but as with everything you use them for, you need to be familiar with the subject so you catch their mistakes.
I’m thinking beginner level so the LLM can support instead of replacing you while you get better.
LLMs do not do translations, they approximate something similar to the original statement in another language. They are very accurate when given a common piece to translate, but wildly accurate when given a sentence which is highly improbable.
If it makes more mistakes than humans and therefor requires humans to check all of their work, and it’s been shown to not be very cost-effective, then what’s the point? Better to just not use the AI at all.
Love that, what a cool dude
Can’t wait for the punchline.
No punchline, just found such hardline requirements potentially entertaining.
What’s wrong with it? Honestly it’s kinda refreshing.
What’s wrong with it?
That’s a complete disregard of a useful tool that brings measurable benefits.
Why are you on fedi if you love the corpo slop machine so much? This platform was specifically made to get away from people like you. Go back to linkedin.
Don’t you think too much of yourself and your role here? Lmao. What a buffoon 😂
Name 1 great struggle that humanity suffers from that “Ai” has resolved.
Shhh, you’re not supposed to notice that on Lemmy.
Nah, be it AI psychosis or anti-AI psychosis - neither should dictate us what to say.
It’s a resolution they’ve arrived after considering the benefits and downsides of that tool. Nukes can be used for efficient large-scale excavation, but people were sane enough to realize why nuking themselves was a horrible idea.
Considering only the benefits and not the risks and drawbacks of anything is not just reckless. It’s zealotry.
Nobody proposed to ignore risks, but okay, you wanted to win this argument - so you had to invent it.
where are these measurable benefits? so far they haven’t really realized anywhere
Authors and maintainers of many open source tools that you use every day - whether that’s curl or linux with its army of developers - openly discuss the value LLMs bring. You either don’t care to listen or you choose to ignore their words because they contradict your beliefs.
Funnily enough, many people in the industry with years of experience and a strong reputation have a fairly balanced view of AI, recognizing both its advantages and disadvantages. Only on social media do you find people who blatantly deny any positive contributions (or folks who believe that AI is irreproachable, but you won’t find them on fedi).
There’s also huge downsides that everyone always seems to forget (enormous power/clean water usage, noise pollution, etc.)
To be fair, this is what bothers me more than the supposedly non-existent benefits, because I’ve seen LLMs being useful as tools, but the impact on environment is not clearly communicated (I’d wager it’s because no one wants to admit that it’s enormous). For that reason I’m trying to use local LLM instead, and it is more limited in usefulness but that’s fine by me as it’s not the only tool I have
Clearly they don’t find it useful.
Up to them. I only answered the question.
Yeah how do they want to enforce this if people use it for brainstorming?
iirc their goal is not to have a perfectly enforceable policy, it’s to cultivate a culture around the language that is fully in favor of human contributors instead of ai. to let people learn and improve as they contribute, instead of simply merging code that looks good enough; they call it contributor poker
It might not be enforceable, but why would they want to contribute to something that’s very against what they use? Out of spite?
Like all of this ranges from unenforceable to spuriously enforceable (eg for rule 1, you can guess whether something has AI vibes—with vibe code it might be easier if the AI has just hallucinated a function or something). Seems more for the purpose of making a point than anything, or perhaps relying on others respecting your policy, but other projects with much more lenient no-AI policies still have people flagrantly breaking them.
Encoder-decoder language models and all sorts of stuff were used for translation and spellcheck, long before “LLM” was in anyone’s vocabulary. Embeddings models were used in documentation searches, in IDEs, and other places. Whenever you used any search engine, pre Sam Altman, you were likely hitting text models too.
They worked alright.
It was not an issue. No one hated them; they are simple tools with a specific function.
I think people need to be careful of spilling (quite reasonable) hate of Tech Bro AI into the wider, older field of machine learning. In spite of the effort to conflate them, they aren’t the same thing.
