I feel like this could have been avoided if Torvalds had allowed himself to use C++, instead of suffering the ritualistic pains and frustrations of C for an entire career.
Anyone would lose their love of the craft after that. AI isn’t how you find it again, and it makes me sad to see. Not sad in a depressed/AI doomerism way, but sad in a “the people I looked up to are old and washed up now” way.
Watching Torvalds actions tearing this place apart is hilarious
Almost anyone can cut and paste a face on a different body using a classic photo editor.
It still takes a profesional to do it in a way that delivers a quality result.
Linus Torvalds is a professional, the linux terminal is his baby, if we can trust anyone on this, its him.
The amateurs using ai to produce slop that resembles what we perceive as quality, is sometimes hard to tell but 25 years ago people though half life 1 looked insanely realistic and that graphics could not possibly look any more real. Over time our brains will adapt and we will be able to tell slop from real work in a heartbeat. I bet most people here can already tell if a mail was written by an llm and not audited by a human with ease.
We didn’t think Half Life 1 looked realistic against reality, just that it was the most realistic graphics we had seen yet.
It’s not like we looked at the game and went that’s indistinguishable from reality.
but isn’t using AI inherently unethical?
Yes, and the people doing mental gymnastics to defend it are unprincipled losers and/or cucks to the billionaires.
The models used to generate code were built by stealing from us. Fuck whatever stupid opinions you might have on copyright; you were exploited. Elon Musk, Mark Zuckerberg, Jeff Bezos, Sam Altman, etc are making money, corrupting our democracy, and destroying our way of life because they stole our work. They wouldn’t be able to do that without the massive theft of labor they’ve committed, or if our government actually enforced the laws that they broke, and are currently still breaking.
Why should these scumbags (or anyone) be able to steal and profit from our labor? Are we going to accept our assigned roles as mere cattle for the megacriminals? Will you be happy when your daughter is conscripted as a “breeder” for Elon Musk? Or when your son dies fighting in a war for resources so we can build more data centers?
Or how about when the subsidized tokens run out, you can’t afford to use the models you’ve become unable to function without, and are forced to go work in an Amazon warehouse or copper mine or some shit because you eagerly and enthusiastically contributed to the monopolization of your own labor, and were too stupid to realize it before it was too late?
My subjective opinion is that yeah it is unethical, but not more then so many other things we normalised.
Systematicly slaughtering animals for meat for example.
I am not a vegitarian by any means. You cannot fight on all fronts at once and have to pick your own battles in context of your own lifes challenges.
Cars poison the air and kill many people, i dont drive a car for that reason because thats a fight i can fight. I have sensativities with food, so i am not cutting out meat. (I aim to do so some day)
I moved to linux some years ago by getting help from an llm which requires datacenters build by malicious companies. I don’t have a google or meta account and my cloud and media is hosted locally on recycled hardware. Where for others those services require datacenters build by malicious companies.
Life is various shades of grey.
I’m worried about cognitive surrender. There’s evidence of people losing skills they once had because they stopped doing it themselves and just get AI to do it for them.
“Using AI for Just 10 Minutes Erodes Your Ability to Persist at Hard Things”: https://vcresearch.berkeley.edu/news/using-ai-just-10-minutes-erodes-your-ability-persist-hard-things
They should run the test with calculators, because they’ll see the same effect. Turns out that giving people a tool to make things trivial, and then taking that tool away, halfway through the goddamn exam, is a lot more frustrating than just asking them to do some algebra on paper.
‘Bicycling for just ten miles makes walking back seem like a pain in the ass.’
going to just copy and paste my reply to your other comment: AI is not a tool it is a replacement, there’s no skill in using AI
If you give up all critical thinking to the LLM, than it is indeed a replacement, but there are lots of ways you can use AI without giving up all your critical thinking.
And it always takes skill to be able to evaluate the results of an LLM.
so you didn’t look at the study, ok
If he’s still auditing and understanding the code. Something that he’s done as a project manager for decades at this point. I don’t see how it will. If the code is generated by a person or a machine. The auditing isn’t much different.
Ironically, I find the LLMs to be better at auditing code than generating it in the first place. Like the humans they trained under, LLMs make mistakes and miss important things in both endeavors, but: when an LLM finds a real problem, that’s a win no matter how you look at it.
Definitely. I just hope that sooner rather than later it’s possible for them all to move over to local open source models. Supporting wasteful content thieves is not sustainable.
That and that all the GPU manufacturers Etc have their sound blaster moment. These days a lot of socs have quite powerful Graphics accelerators on them. At some point, someone’s going to release a bit of silicon equivalent to the AC 97. May not have been able to compete with sound blaster on the ridiculously high end of the market. But was good enough for 99% of tasks. So that most people don’t even remember sound blaster anymore. Even though they still technically exist.
I have been using “onboard graphics” professionally in our products, including products which do 3D rendering, since 2014.
I think GPUs have a little longer life than sound blaster cards still in them, massive arrays of relatively simple CPUs have more purposes than just graphics rendering.
I searched what sound blaster is, but I don’t really understand it. Are they a company that made audio cards in the ancient times? I think I remember some people said that having audio cards was not standard.
Their website seems to have weird hardware targeted at gamers.
Maybe they sell audio interfaces? But I’m confused why gamers of all people would be the target group. I got an audio interface for my guitar myself and some MIDI things.
Back in the DOS days and early windows days. They were one of the gold standards for PC audio. Even if you didn’t own a sound blaster card, you likely owned a clone that emulated one. They were the bees knees so to speak when it came to PC gaming.
But by the late 90s the company had spread itself so thin on branding and other things. Sound blaster CDROM drives? Yep. Sound blaster video cards? Yep. Etc Etc Etc. They tried to parlay being a lucky defacto standard into something else. Wanted to be a premium brand. Prices go up, quality goes down. Kinda like Nvidia now. Fake frames, AI slop, shitting on traditional customers.
Turn of the century and a standard set in 97 finally started rolling out on consumer motherboards. No longer when you bought your computer if you wanted it to make anything more than basic beeps boops and buzzes. Did you have to go out and spend hundreds or thousands of dollars to get an ad in card. It was on board. On the motherboard.
The newest computer I own is an early ryzen from the previous Dark Times. Not to be confused with these dark times. So I don’t have first-hand experience. But it is my understanding that Intel integrated Graphics isn’t quite the joke is that it used to be. Just a bastard stepchild that Intel can’t decide if they really want to support of not. And some AMD chips have an APU on it which is similar. But still chokes under anything very intense. It’s getting close. But just not quite. Give us a truly standardized setup with the basic power of a mid 2010s graphics accelerator and vulkan api, most people would be in business.
Honestly I’m hoping something along this line happens and becomes common with risc-v systems. Like Mali with arm.
Thank you for the explanation.
I have a laptop from 2022 with an Intel CPU. It’s pretty good to what laptops did last decade. You can play Factorio, Minecraft and RimWorld for example in pretty reasonable graphics. It’s not that great of course, but it works fine, especially if my laptop didn’t have heating issues (framework 12)
yeah I agree with this statement. it’s not that LLM are necessarily bad it’s making people reliant and lazy which leads to complacency.
it’s making people reliant and lazy which leads to complacency.
Pretty much the whole point of every invention in the industrial revolution.
Absolutely not. Lots of thing were created to make jobs faster and more importantly faster.
Take the simple washing machine for example. Early iteration, before electric washing machine, caused a lot of injuries because people would get their arm stuck in the machine’s mechanism.
So by creating the modern washing machine, we freed lot of time for people to do less menial stuff and made their life safer by not having your arm pushing clothes through a dangerous machine.
Take the simple washing machine for example
or the cotton gin, or the steam engine…
Now one wo/man + machine can do the work of ten wo/men… and 9 wo/men were freed up to “do less menial stuff” - and not get paid (or enslaved) for it.
Only in the sense using a calculator makes you worse at long division. You’re still doing math.
Every pearl-clutching article like ‘doctors got worse at spotting cancer when we gave them a tool and then took it away’ is only measuring before and after. None of them seem interested in asking: how much more cancer did they find, with the tool?
AI is not a tool it is a replacement, there’s no skill in using AI
Weird how software engineers are better at ‘being replaced’ than novices. They seem to get better results by knowing what it’s supposed to be doing.
What a funny coincidence.
Linus explained in that Git tag message that he “did what I preach” and used modern LLM tools to make sense of patches that were hard for him to follow.
It’s already too late. He’s gone down the rabbit hole.
I can vouch for that: As a programmer, I can generally see both programming AI and general LLM use if it was copy-pasted. I know a guy who isn’t a programmer who’ve used Claude Code to make an app, and things are working, but it’s sooo clear even to a junior SE that it’s vibe-coded. There’s a clear lack of consistency, architecture, maintainability, and general human readability.
However, if the code/text was reviewed and edited by a capable human programmer, it’s incredibly difficult to single out what is a product of AI and not.Actual programmers who understand and can write code themselves who use AI simply as a tool, NOT a shortcut, will produce indistinguishable code. Linus is an incredibly smart and capable guy, and for specific uses, I’m confident in him using this tool responsibly.
(The ethics of AI use is a different topic however…)
How do you see the longevity of it. You obviously learnt how to code before ai. What about those learning now that are learning only having access to ai. Id anticipate the quality over time will deteriorate
I think we’ll unfortunately see a degradation over time of capable programmers, as it’s with most other things: Humans learn poorly from seeing things being done for them, rather than doing it themselves. AI can be a good resource for iteration and a sort of rubber duck, and even for explaining things (with external confirmation). But as it currently stands, people find it much more attractive to just let AI do the work for them. That’s the most prominent work culture after all, based on capitalistic values: Economic growth above all.
I see it in schools as well, unfortunately: Students find it much more appealing to get good grades than to actually learn. Not everyone, of course, but still way too many.
And I do understand many of the students: The bar for junior programmers has now increased, as a lot of employers simply think “why hire a human that is on the same level as AI?”.
Students find it much more appealing to get good grades than to actually learn.
Because there is a direct impact on your chances for a better life. Having better grades give you access to better schools (and in the US, scholarships as well).
The emphasis of our school system is results.
This message explains the difference between being a centaur or a reverse centaur. Pluralistic explains it in more depth.
Meh, its less exciting whenever Claude Code is involved; I am wondering how many undetected bugs will sneak into the kernel. If he was going to use an LLM, he could at least use something local and not connected to one of the many nasty big corps that are ruining shit.
You wouldn’t want him using a local model unless he was running some fancy rig that could run a high end frontier model. He might not have such a thing laying around. Using the best out there for something like this is how you help avoid bugs.
He gets paid more than enough to afford high end hardware.
He definitely can afford it yes, but its still a big ask to want someone to spend 15-25k or whatever it would be (shits getting more expensive by the day) to use a tool that he can just do pay per use. The code is open source so its not like he’d be using a local llm to keep it private.
Wendell from Level1Techs built Greg KH a monster rig for free, I’m sure they’d be willing to build one for Linus too.
Dang, wish someone would build me one for free! Thats super cool of them though!
-
AMD Ryzen Threadripper 9960X
-
GIGABYTE TRX50 AERO D Motherboard
-
Samsung SSD 9100 PRO 2TB SSD
-
Noctua NH-U14S TR5-SP6 Cooler
-
Intel Arc B580 GPU
-
Fractal Design Torrent E-ATX Case
-
Seasonic PRIME TX-1600 1600W 80+
-
Titanium PSU
-
ASUS ProArt Display PA32QCV 31.5" 6K
-
HDR Monitor
-
Honestly, his mind and the community is probably one of the best things out there at the moment. They live the Linux Kernel, while that thing he used does not think or understand anything. Eh, it would be better if they moved with care instead, as that is a good way to avoid bugs too. I don’t have a lot of faith in gassed up LLMs.
He’s doing it though. So that’s kind or moot. Fable is out there finding bugs in these things, so seems like if he is going to do it, using the thing that’s already making it more secure by finding multiple exploits is the right choice of what to use.
Fable is out there finding bugs in these things
Ehh, that’s debatable. An LLM will gladly spit out a vulnerability report, but as LLMs lack any understanding or reasoning capability, there’s literally nothing guaranteeing that it has any semblance in fact or reality. They are just random text generators after all.
Going by Anthropic’s own numbers for things like Mythos, it “found” 29,439 vulnerabilities, when a human actually checked if the reports made any sense (ignoring validity) that dropped to around 6,200 reports, when they were tested for validity that further dropped to around 1,300 reports, and when it came time to actually patch the issues only 516 made the cut (i.e. the rest might have been issues, but weren’t considered exploitable, etc.).
516 out of 29,439 is a 1.75% success rate, not exactly groundbreaking.
Now to be fair, according to Greg KH at a recent talk, Mythos is seemingly improving. Of 79 vulnerabilities it reported, a solid 4 were real. That’s a 5% success rate, though it performed much better when it simply reported other developers bug fixes, something it did 11 times.
And how many security vulnerabilities have you found? 516 is still a lot. One of those had been in the Linux kernal for decades.
Edit: and many of those real ones from the 6000 group anthropic says the maintainers just didn’t want to fix as it was outside their threat model. A vulnerability in the Linux kernal, is different than one on an alarm clock app or something less serious.
I am wondering how many undetected bugs will sneak into the kernel.
Less with LLMs than without. https://projectdiscovery.io/blog/the-vulnerability-curve-bent-with-the-ai-curve https://www.darpa.mil/news/2025/aixcc-results
Yeah, but you are still relying on models that are going to go up in smoke when the bubble fucking crashes. As people are rightfully hesitant to pay the real price for these things, these things are controlled by mega-corps which is ultimately a bad thing because a centralized authority sets the terms of engagement with their services. They are the ones actively scraping open source projects so much that it acts as a DDoS attack, hovering up personal data of other people as if it were candy. Not to mention the other effects it is having on jobs, AI systems being central to massive data breaches, etc. A few positives don’t outweigh the massive negatives that come with the technology.
If the models all go up in smoke, we still have the code (and its bugs) - so let’s get as many bugs out as possible as soon as possible and not accept a bunch of garbage code - as has always been the case, it’s just happening faster now.
Trainers (all over the world) are behaving badly, because they can get away with it. I’m not sure where that ends up, but it doesn’t change the generated code or the discovered vulnerabilities or the fixes.
The personal data has been abused for decades already, but again - now it’s easier for more actors to abuse your personal data and mine. Maybe, someday, we’ll get serious about enforcing privacy policies and not just have completely toothless theater about it.
Yeah, personal data has in fact been abused for years, it’s just getting worse though…With all the latest breaches that have leaked a lot of shit. It’s why I am constantly fighting all the bad bills whenever they are being talked about, making sure I am a pest to my representatives. Making sure my voice and opinion are loud and clear, also making sure others speak up too. That makes a difference. We just have to fight harder for more bills with teeth and consequences for the various entities that handle data and let it spill. Like actual business ending fines and real punishments. Setting actual standards that are sane, secure, can be optimized and built upon.
That code is still suspect to me, as random code generators aren’t known for creating maintainable code…People have to audit that shit to make it serviceable, which greats bigger workloads. Breaking the spirit of mostly volunteers with such inexpertly crafted code is just not the best idea.
Yeah, Trainers do act badly because they weren’t regulated to hell and back early on…With clear and decisive lines that shouldn’t be crossed. With rules that needed to be followed, people paid for their data or compensated for hammering personal websites. A lot of the data gets data laundered and put in training sets, but can be found in models, said models can be made to reproduce phrases. It’s honestly nasty work, that should’ve never been allowed to get out of control like it has. However, it was doomed to be this way because an absolute morally compromised fool is sitting in the White House so the major US creatures…Are going to get away with a lot for as long as possible. As for other countries, well, they don’t seem too keen on regulating so called “AI” as much as they should be. They are coo coo for corporate lies.
I am a pest to my representatives
Keep it up. I’m in Floriduh so talking to our representatives is more or less like talking to a brick wall. They may accumulate counts of how many people raise an issue, but for the most part they’re serving the federal administration’s agenda, not listening to their constituents.
People have to audit that shit to make it serviceable, which greats bigger workloads
I mostly ignored LLM code generation until around August of 2025 - it was still pretty raw back then, but it was showing potential for significant improvement, and since then there has been significant improvement. At this point, I find that if I ride herd on the LLM agent and keep it focused, structured, addressing technical debt frequently, etc. I can actually review the generated code more quickly and easily than similarly mature code by human developers. What is coming out of models like Opus 5.5 now is looking like what I would write if I had all the time in the world to cover all the edge cases, handle all the errors, do the logging and auditing and all the other “good stuff”. It does explode the code for what was a simple “happy path” thing into a much larger structure that takes the real world problems into account and gets in front of them, handling them gracefully and presenting clear messages about what went wrong when unavoidable problems happen.
I guess what I’m saying is: yeah, we’re still getting more code to review, but it’s getting easier to review and the presence of meaningful automated tests and consistent documentation and non-rotten traceability from requirements through all that is actually resulting in lower overall workloads once you look at the big picture of what it takes to maintain code in the real world with security and other issues and not just the happy insecure proof of concept and ship it world.
Breaking the spirit of mostly volunteers with such inexpertly crafted code is just not the best idea.
Yeah, if your models are giving you inexpertly crafted code, you need to keep iterating with them until it’s expertly crafted before shipping it, don’t just throw up a pull request because some bogus tests passed. And, there’s still a difference among models - they’ve all improved rather dramatically over the last 12 months, but for my purposes (coding robust, secure, safe applications) Claude’s Opus and Sonnet are still “in the lead” by a pretty good margin. Other models are better at other things…
Trainers do act badly because they weren’t regulated to hell and back
There’s not enough transparency or accountability in the global business landscape for any kind of regulation to actually work. If these regulations are going to have any chance, the first step has to be true transparency - like FOSS does with open source development repositories, and more.
people paid for their data or compensated for hammering personal websites.
Keep dreaming. It would be a wonderful world, but it’s far from where we are.
said models can be made to reproduce phrases
Like English majors can quote Keats after graduation… I’m not sure that’s proof of plagarism in their other writing.
an absolute morally compromised fool
Yeah, it’s been an interesting ride this cycle - seems like they’re trying to tear down the government institutions in preparation for something like Palpatine disbanding the Senate… destroy all confidence by blowing up every problem of the past and flaunting it as something actively done at the highest levels. If they succeed in holding power through the upcoming mid-terms the ride is only going to get more interesting…
I’d definitely like someone to champion an open-weights model under 32 GB. But part of that desire is - I’m not sure they’re there, yet. Local models are only a few months behind, but hoo boy, the last few months have been big.
Coding models could be tiny, but distilled versions keep preserving pop-culture knowledge and world history, when reasoning through C++ requires essentially none of that. It needs the tokenization to recognize why a comment begins “Jesus Christ.” It’s fine for “treat the hash like Lot’s wife” to go completely over its head. Keep the weights that know a regex for e-mails, not the capitol of Florida. If it speaks a dead language I want it to be Algol.
“Improvement” huh
?
I too love
filling my code with things that silently reduce my understandingimproving my code usingstochastic prediction models trained on tainted data filled with hidden injection prompt attacksAIwhat’s the worst that could happen? /s
I’d love to see a pre-AI fork of the Linux kernel.
Be the change you seek in the world.
I think some of the BSDs have no AI clauses.
I ran it all through Slopscore and consulted open-slopware, and here are the results:
- ⬛ Linux has at least 2,274 slop commits, to date.
- 🟧 FreeBSD has at least 53 slop commits, including some from Anthropic themselves.
- 🟨 OpenBSD and NetBSD have a couple of slop commits, but nothing colossal. NetBSD also requires any AI-generated code to be signed off by the core maintainers.
- 🟨 DragonFly BSD has at least three slop commits.
- 🟥 MidnightBSD has at least 292 slop commits.
- ⬛ NextBSD is entirely vibecoded.
- 🟩 MirBSD has an anti-AI policy, but only supports i386 and SPARC, and is currently unusable anyway.
- 🟩 HyperbolaBSD has an anti-AI policy, but hasn’t been released yet (they’re still a Linux-libre distro).
- 🟩 CoffeeBSD is a slop-free fork of FreeBSD, but has had no commits since it was forked.
I guess that settles it you’ll never have a decent kernel without AI code.
Idk about that. Hyperbola’s BSD kernel is looking good, and Haiku is coming along nicely.
Redox OS also has an Anti-AI policy, I believe.
It does! Also, 9front and ironclad have anti-AI policies, and illumos is untainted (though does not mention anything about AI in its CoC).
You can do it always
Okay, that was always an option.
I can see people genuinely switching to NetBSD because of this.
deleted by creator
