I actually disagree with a lot of people here.
If you allow llm to infect code, you will have no programmers that can code once your pre-ai ones (like linus) phase out.
It’s not really ai that is the problem but the skills it eliminates. You can’t properly fix bad ai code if you don’t do code without ai, you will not gain the experience of decades of manually coding properly.
The most experienced coders that exist today are the equivalent of artisanal stone carvers and wood workers who built some of the most beautiful buildings that exist of older times. Modern construction methods phased them out and now the number of people with their skill is tiny worldwide compared to thousands of them in every country in the past.
The tech industry is going to eat itself and won’t realise until it’s too late.
Okay but 99% of corporate codebases are already insane crap. The LLMs write boring, overly safe code, but at least they don’t make obvious IDOR bugs like my human coworkers.
I’ve seen LLMs put sensitive secrets in git just like any incompetent junior dev
Harness + got hooks + formal verification trivially prevents everything like that. You’re just holding it wrong.
Yeah but when you’re gone nobody is going to even be able to tell the difference between one set of code or the other. Instead the competent programmers will be replaced with tech priests preaching to the machine.
Embracing this technology eliminates competency. Consider how the art industry is currently refusing to embrace it and instead fighting it, what does this do? It ensures that the future has artists who remain competent and does not turn the field of art into a bunch of people who give prompts to an AI.
To really understand this you need to imagine a future where the art field embraced AI and all manual art stops, instead being replaced by people that give prompts to AI to make art. This is the future of the programming field by embracing AI. All future programmers when the manual programmers phase out will just be people that only know how to prompt.
It is incredibly shortsighted and only the realisation that everyone in the field will become like ai “artists” once the current experience phases out really makes this clear. It will suffer a total skill collapse.
But the art field is making final products that have no formal correctness. The programming field is making infrastructure that never gets seen by end users, and we do have formal verification. Two completely different things.
The skill collapse already happened after the 90s which is why all the software is so shit right now. Now the competent people can take over again and rewrite everything in immutable functional languages with complete unit test coverage and viral copyleft licenses.
rewrite everything in immutable functional languages with complete unit test coverage and viral copyleft licenses.
Cool! Where is the git repo?
Im still working on it but you can use Pharo for now
This is why anyone who uses llms should know how to code and they should review the code.
People don’t exactly stay on top of their game by reviewing slop code.
You don’t learn that stuff get your degree and remain an expert for the rest of your life.
Yeah but if they don’t write code they won’t learn. A lot of coding isn’t really like learning to drive a car and once you know how you’re done, it’s a persistent lifelong accumulation of experience in architecture.
The field absolutely will suffer a skill collapse when the existing generations of experienced pre-ai programmers phase out. There is no replacement mechanism in the system and the industry is young enough to have never even had to think about this problem ever occurring.
I’m not pro-LLM but this already happens all the time at corporate levels when the bean counters decide to fire (usually constructively dismiss) their senior talent. They have a complex codebase that captures domain specific logic well and uses good practices but then they try to have a dev team that’s half the size and 1/10th as experienced try and fail to maintain it. Then customers complain and it gets poorly patched over and over again until something important breaks and they have to stop touching entire parts of the stack or they leak a db dump or get ransomed or the site goes from 99.9999% uptime to 90%.
I do understand that at an industry level you’re describing something like the memes where only millennials can edit a pdf because the boomers/gen x never tried to learn tech and the zoomers were never exposed to non-mobile-app workflows as kids. Where the skill itself is mostly generationally bound. But I do think the skillet will remain, just not at the corporate shovelware factories where it was already actively despised as a cost center to be done away with at the earliest opportunity. Because they already deskill there! And outsource with zero investigation of competency (not a dig at workers getting the outsourced jobs, even the ones that don’t know their stuff, because hell yeah I salute that).
Makes me think it is also representative of the trend in physical production and construction. In the US there were decades of high output. Starting nineteenth century, even more after WW2. But then everything became complacent, outsourced, unmaintained, deregulated.
It is a general trend for sure, not a new innovation. Like elon musk doging the government. Same effect. Now all that skill is gone. Did you see the post that FBI now can now hire people with criminal records of bestiality?
All the stuff about Iran vs US military production and capacity also. Iran was exempted from the trend due to circumstances.
Maybe they try to eliminate the middle class as much as possible so it will not be an annoyance for them in the future.
Ok but if you put it like that it doesn’t sound like a bad thing.
Until you realise that software is in everything from government services, to hospitals, to the means of communication
A lot of that is still paper records here.
Or COBOL that nobody dares to touch anyway.
But like, for job security
Surely in 50 years it would be ok if the only people that hand wrote C/C++/rust were Amish or working on an obscure hobby.
Like sure everything sucks now and will suck in 5 years, but remember cold fusion only took 50 years.
I really dislike how everything this man says or does is interpreted like it is instruction from god.
If Linus torvalds jumped off a bridge, half the Linux nerds would line up to do the same.
Yes, the fucker said some pretty fucked up things.
I can’t wait for no one to know how any of our software works in ten years, why it’s so bloated, or why it breaks all the time.
That’s pretty much how a large part of our software is today. It’s only going to get so much worse…
Even the TUI applications are now made with reactjs for some reason.
Reject slop, return to teletype
I’ve been in software for nearly two decades. The industry has fundamentally changed with the advent of language model tools. I don’t like it, but Pandora’s box is wide open and I want to understand what the path forward is.
I don’t write code for work anymore, I babysit a language model. Maintaining a sense of code ownership and comprehension is a real risk, and we are pushed to use tools that turn our brains into pudding.
Is this what the old timers thought when fancy pants compilers showed up? The reason I disagree with this analogy is that the new tooling is fundamentally incompatible with software freedom that cannot be solved with gcc et al. I can’t afford a $10k computer to run local models.
The idea is you’re basically expected to be a manager so the LLM is now your junior engineer except you don’t get the joy of building someone up and watching them grow you get an expensive robot that never learns and can’t remember what you said yesterday.
I really do miss mentoring junior engineers. At least they never rambled on and on about “load bearing seams”
“You are absolutely right!”
Two of my coworkers are constantly adding to and tweaking our repo’s skills folders. I guess that’s the equivalent of mentoring for them and it actually has made the responses significantly higher quality.
It’s pretty funny watching software go from arguing about functional programming, total test coverage, side effect free code, and mathematical proofs of correctness to “eh just sorta run your eyes over code you didn’t write and hope that 6 hours into your shift you’re not letting anything critical by”
Erlang, Haskell, my beloved. I will still love you while the others are lost to a storm of madness.
It’s not funny at all. I’m gonna develop male pattern baldness from pulling my hair out every time I have to interact with LLM-using ‘programmers’.
This is why I chose to laugh. I am making games as a hobby now, there’s nothing left in computer touching for money that I used to love.
The reason I disagree with this analogy is that the new tooling is fundamentally incompatible with software freedom that cannot be solved with gcc et al. I can’t afford a $10k computer to run local models.
That’s actually a good point but if we consider history the barrier to entry to software engineering was huge 40-50 years ago, not many people could afford to do it locally. Then again it’s different because software as a whole wasn’t really that important 50 years ago.
Then again “software freedom” as we know it may just as well be a coincidental transient phenomenon because capitalist markets arranged themselves in such a way to make the hardware and software required to build and maintain software accessible enough to the general public, that “right” can conceivably be taken away by the same system again, there is no reason to assume it’s here to stay.
Honestly I’m just rambling at this point I have no idea where any of this leads I’m pretty much hoping the whole thing is gonna be a dud and we’ll return to the same old way of doing things, capitalism is a fuck, death to america.
I can’t afford a $10k computer to run local models.
Tbh you don’t really need to either. With models like Qwen 3.6 35B A3B (which is quite close to the performance of frontier models from a year ago) enough of it fits onto a 8GB GPU - with the rest sitting in RAM or swapped out - to have it running at a speed that is very workable with. On my work laptop with a 4060 equivalent and llama.cpp + CUDA I get around 15-20 t/s for coding tasks. The equivalent Qwen 3.8 variant should be even better once that releases. I strongly believe small models like these running locally is going to be the way to go once the bubble bursts.
I can’t afford a $10k computer to run local models.
Hopefully that is only temporary.
I imagine few people could afford a Unix system in 1987 when Stallman first wrote GCC.
It’s not magic these models are big and you can’t just squeeze them onto smaller devices. Quantisation isn’t magic either it’s literally reducing the accuracy of a larger model to fit on a smaller device reducing its performance. If the data doesn’t exist in the model weights then the model will perform worse and you can only compress data so much. The only meaningful tech shift in the space is unified memory but it’s not going to be enough for anything this is fundamentally going to need silly quantities of ram for any effective model to run.
The main limitation is just RAM on the GPUs. Unfortunately the bubble itself drove up prices on that exact thing dramatically but this not a particularly expensive thing normally.
normally.
GPU prices shot up in ~2018 and 2020 and [etc] and never came back down. Nowadays a 9060XT (the useful version, mind) costs on its own what a decent entry level PC used to. We are never going back to normal.
ⓘ Remember to enable JPEG-XL support in your browser, even on your phone!
Excuse my ma’am but let me tell you about a little thing called supply and demand. smirks insufferably just before getting punched
We’re at a really different point technologically and given the whole thing of Google’s paper and the architecture is more data more layers more good there is a huge incentive to run on as much ram as possible which is expensive.
We’re not going from hand soldered resistors to nanoscale transistors again. We’re looking at some serious physics walls. Even if photonic chips leave the lab.
this year 1 CS student used her brain to fix a Linux bug, and now thinking critics on Linux face an uncomfortable choice
I don’t think it changes much of anything if linux doesn’t crash, Linus already said he will allow llm based code if it has been fully reviewed and he already said he considers it a neutral tool which is good in some instances, the fact he used it for troubleshooting isn’t that surprising or groundbreaking tbh from what was already public knowledge.
I don’t really care that much about whether the code I use has been made with LLMs. I don’t like working with LLMs and I don’t like working with people who heavily rely on LLMs. And since this is a matter of me volunteering my time it’s an easy decision to distance myself from projects with an everything goes LLM policy.
the linux kernel is compromised? install openbsd immediately?
ⓘ Remember to enable JPEG-XL support in your browser, even on your phone!
I’m being baited >w<
Hi :3
I knew it when I saw the floppas!!!
I still recall you saying about OpenBSD and so when I have ample time I’m gonna install it again and Cause Problem™
ⓘ Remember to enable JPEG-XL support in your browser, even on your phone!
Alright everybody! It was a good run. This time let’s account for the Y2K issue from the start, and I’m thinking we try trinary again, I think we can do it this time.
Writing correct software, especially at a lower level, is like 10x harder than people think. It’s always been hard, and now llms can actually fix the obscure concurrency and memory safety bugs. I’m all for it.
writing correct software
LLMs
The problem is LLMs always sound very confident when “fixing” things and will very rarely report that it’s out of its depth and refuse to work, and yeah sometimes they genuinely just do good but also unless you’re really vigilant it’s easily possible it’s gonna introduce more bugs than it fixed.
I’m not saying Torvalds is stupid enough to just uncritically let this thing loose on the kernel but I understand the trepidation of open source maintainers about this stuff.
LLMs will actually double down on incorrectness and make even wilder mistakes unless you bully them until they go, “I admit I don’t know how to do it”. Unless a reviewer really knows their stuff they can easily fall victim to trusting a bullshitting LLM.
This is on point. You’ve got to keep pushing back until it realizes some overlooked assumptions and shit like that. I cope by architecting my codebase in a way where I know I have guardrails in place for bad LLM guesses
It doesn’t realise anything. It just generates a single token, appends that to its input, outputs that, ends, and starts again with the new input.
The anthropomorphising muddies discussion.
How do you know you’re not just generating the next token and appending it to your input?
Nah just kidding I kinda tend to agree because techbros are already starting to speak about LLMs like they’re actual conscious beings and it’s getting super weird but saying stuff like “the text predictor that is used as a simulacrum of human reasoning is deficient in such and such ways” is kind of a mouthful and also most non-technical people are gonna stare through you if you talk to them like that.
It’s pretty easy to change “keep pushing back till it realises” to “adjust the prompt to bias outputs towards” and similar.
:feedback loop intensifies:
now llms can actually fix the obscure concurrency and memory safety bugs
Lol. Llmao.
Here’s one of the few examples where I’ve seen it be actually used in an okay way.
Full transparency, used to generate a standard check and not do giant thousand line PRs, honesty about false positives, and no bloated claims really.
LLM is a statistical model, it does not reason. It is not capable of neither honesty nor deliberate lies. It works sometimes, because it was trained on thousand standard case analyses by human programmers, but it is also prone to hallucinations, which are unavoidable by design.
Well if you’re going to be pedantic, they don’t hallucinate either. They just repeat patterns, and sometimes false patterns can emerge in large data sets.
Well thats not always why they give an incorrect prediction. It’s also because statistics are inherently never going to reflect reality perfectly (“all models are wrong, some are useful”), but also we intentionally inject randomness into LLMs and other ML programs to ensure the output isn’t over fitting and spitting out raw training data. Usually adjustable via seed values, temperature, etc
I know, the reason I shared that example is because there is no humanizing the LLM and it’s used exclusively for its ability to pretty effectively parse syntax to find common higher level bugs in specific C frameworks.
For context: I never use LLMs when programming because I find them to be more of a hassle. They show me down and muddy the water on design principles.
Even a “bad” design pattern that you’re consistent with is 1000x better than a “good” design pattern that is inconsistently applied. LLMs are fantastic at the later. Things that look great in isolation, but begin to fall apart when they come up against the weight of a larger project.
Especially when it’s a new project without a ton of existing code for it to kinda copy/paste.
Using one to find minor use after frees that are almost impossible to hit, or duplicated calls is fine by me though. Those are usually a pain to find anyways, and there’s almost always 100 more important things to spend time on. If someone wants to burn tokens creating a backlog of easily fixed, reproducible errors like that all while disclosing exactly what they’re using to find them sure. That’s not annoying at all, especially when they only open the PRs after discussion with the project maintainers.
All the code is “in there” – it’s a matter of forcing it to produce correct code. There’s tremendous profit if you can.
The same applies to a random number generator.
