Don’t compilers have some guarantee of behaviours? I.e even if you don’t know exactly what instructions and order the compiler generates there is a guarantee that it is computationally/mathematically equivalent to the statements you wrote. Otherwise it would be a pretty terrible compiler. That’s a pretty big jump to writing instructions you don’t understand, since LLMs can’t make the same guarantees about a prompt/request.
Yes, they are deterministic. LLMs are not.
Tell me you weren’t around for the early days of compiler development without telling me you weren’t around for the early days of compiler development…
(Actually, for niche architectures LLVM is also pretty bug-riddled. Maybe with AI we can finally fix some of that.)
Comparing AI to traditional tools is a bad faith argument and it’s a bummer to see Linus fall for it.
I don’t know… tools made by companies who can’t wait to get in bed with the Pentagon to kill brown people around the world would seem to be toxic af.
Surely the Linux kernel gets used by those same organizations too, in some way
Don’t call me Shirley, though
If we’re playing the altruism olympics game, everyone is already throwing stones in a glass house as it is, because virtually all electronics and batteries on Earth are made from materials mined by forced/child labor.
Yes, but… here comes the next squeeze. I know that they would sell us the rope to bang them with, but god damn.
I love how people with a fraction of his development experience are shitting on him now because his opinion on an aspect of development doesn’t align with theirs.
This is a seriously weak and narrow argument. First off, the parallel drawn with existing tools is shallow. When was the last time an editor or a compiler promised to replace employees? Which of those notably accelerates climate catastrophe? Which one of those is the biggest theft of labor in the history of humanity? Everything is political. Externalizing the social and environmental cost of the major LLMs is taking a clear stance in favour of the late stage capitalistic tech empires currently taking shape.
You can’t really control the use of tools though, that is the problem. AI detection doesn’t work and leads to a lot of false positives. The only thing you can do is check if the product is good.
AI is horrible for the world but it is impossible to ban it for coding, as you can easily submit AI generated code that looks human.
He is not comparing the effects that tools have but that you can’t control if someone uses them or not.
The only way I see for controlling the way AI is used, is if the AI companies put restrictions on it.
If you ban AI, the vibe coders will just lie about the origin of their code.
You can’t really control the use of tools though
You start on a false premise. Immense swathes of the legal system are designed to control the use of tools, are they not? Legality aside, most people aren’t sociopaths, and can handle being responsible humans. You don’t even mention that, in many cases, you can just ask.
He is not comparing the effects that tools have but that you can’t control if someone uses them or not.
I don’t agree that he’s talking about verifiability or honesty.
I don’t see the argument that if some bad actors are going to lie in PRs, that we should just let the floodgates open.
You’re correct but still completely missing the point.
The only way to implement anti-LLM rules is to detect AI-assisted code. That, in and of itself, is a problem because it’s impossible to detect good code that was AI assisted. And you don’t need rules against slop because it’s bad code.
If I’m following it correctly, Linus is advocating for tightening submission requirements before it gets to human review, which directly targets the harms FOSS projects are having under the deluge of slop: it raises the bar so that the slop needs to be good enough to warrant a review, which filters out must of the noise, leaving more of the signal.
Anti-LLM policies are like having closed-book take-home exams: you’re left wildly guessing what’s cheating (or not). Instead, make the exam open book, but expect better answers.
Linus also says AI is a useful tool, and he is an expert, so we must accept that he feels that way. I don’t know if he’s commented on the environmental/social/political consequences of AI hyperscalers, but that’s a separate argument.
The tech is here. It’s impossible to detect. It’s foolish to think you can ban it’s use
And it can be run locally, so there’s no stopping it, even if there was political will to do so. (And with a reasonable economic and ecological cost, no less—ignoring IP/political consequences here, since that’s a whole other topic.)
You’re correct but still completely missing the point.
Thanks ;)
The only way to implement anti-LLM rules is to detect AI-assisted code.
From direct experience, this just isn’t true. I contribute to a small open source project, and we’re doing well enough for our purposes by simply saying “no AI”. We’re not checking, we work on good faith. That’s enough for us. We have no reason to think people are circumventing it.
I think AI needs to be regulated on a larger scale. The tools need to be restricted if you want to stop people from using them.
Putting up rules against tool use without making access to the tool more difficult are worthless.
Putting up slop PRs doesn’t have to be intentional from the authors side.
How would you check commits for AI use if it is not mentioned anywhere? How would you enforce anti-AI rules?
I contribute to a small project, where we simply ask that PRs don’t include AI, then we proceed on the basis of good faith.
On the small scale the good faith approach works. But on the scale of the linux kernel it would have minimal to no effect to ban AI contributions.
For the code quality it’s better to enforce proper reviewing and checking of the code instead of banning specific methods of producing code.
Banning AI is as enforceable as banning vscode.
I myself don’t use ai in general and I despise the attitude towards ai of some people in the community but I don’t think banning ai on this side of the problem has any effect on the code quality and the way things are contributed. I think ai needs to be regulated on a government level. Like removing datacenters or financing of ai companies, which both can also be done on a local government level.
I can well imagine things are different at that scale
The last paragraph of the screenshot is problematic regardless though.
Strongly agree with this, except for “everything is political”, which I firmly disagree with.
Linus seems to be tunnel vision in focusing on how good it is as a tool. That, relatively speaking, I couldn’t care less about. That tunnel vision, however, is existentially dangerous. He is, in some ways, acting as bad as the likes of Thiel.
Lol, I could kind of understand (not really support) his use of AI. While I doubt it’s something you can do for long without starting to make compromises, people who were a really good programmers before will probably use it responsibly for a while, until their skills inevitably erode and (more importantly) their willpower slowly runs out.
But this argument made me loose all my respect for him. It’s so utterly stupid. I hate it when i have to check my assembly code line by line, because it’s not guaranteed that the compiler will compile the same code I’ve written deterministically, and basically just randomizes what it outputs. Or when my editor does not save what I typed, and instead decides based on a random language statistics what to actually save, so I have to double-check my text word-by-word every time.
Oh, wait. No, because the tools are deterministic, and that doesn’t happen. But of course, adding randomness into your workflow, where you have to focus and review everything line-by-line, is the same as using deterministic tools, lol.
where you have to focus and review everything line-by-line
Do you not think all code deserves this level of focus?
people who were a really good programmers before will probably use it responsibly for a while, until their skills inevitably erode
Damn, dude, I had mad 6502 assembly skillz in 1985, I could even write it out by hand - pencil on paper, and those slowly eroded through use of compilers for higher level languages until my last assembly experience in 1996, today… I’d be learning everyhing specific to assembly code from scratch. I guess I’m just useless now.
the people I respect and call intelligent are not the ones that shy away from using new and interesting tools.
Gotta agree there. We had two clear camps of programmers before LLM agents came along, and the camp I least enjoyed reviewing the code of before (and now) of is the camp that resisted adoption of new interesting tools the most loudly.
“Deterministic” compilers and reproducible builds are actually fairly difficult problems. (Yes I know technically the compiler is deterministic, but AI is also technically deterministic since you can set the seed).
It’s not just about being deterministic. A compiler crucially follows clearly defined formal rules on what is a correct output given some input, which the programmer can fully rely on.
which the programmer can fully rely on.
Most of the time, today. Mid 90s I was constantly checking compiler output for bugs, and finding them. Early 90s the compiler bugs were finding me, particularly in Borland Turbo C++. I turned off the linker optimizer in 1992 and couldn’t turn it back on until 2008 without getting difficult to identify bugs.
There is a clear line between deterministic code and what comes out of an LLM. One of the bigger problems there is that LLMs have a rather large “context window” they operate from, and human operators don’t readily “grok” how the content of that context window enhances and degrades the desirability of the outputs you get.
I almost exclusively use LLMs to generate, review and test deterministic code, then use that code for “the real tasks” with zero LLM involvement. The output of LLMs is still unpredictable, but once they’ve “said” something valuable, recognizing and using that value is a deterministic activity.
You can add deterministic testing to AI output though?
Wouldn’t it be just… writing code yourself? It’s just instead of writing what you want directly you are writing tests for it.
instead of writing what you want directly you are writing tests for it.
LLMs write pretty good requirements, specifications, code, and automated tests from chat sessions. You do need to keep an eye on what it’s writing to make sure it’s not hallucinating, but hallucinations have been reducing in practice rather dramatically across the last 12-18 months of model development. They still happen, you will always need to guard against them, but usually a quick read of what is written followed by a quick test of the product will find the worst offenders. In the end, you need to check the final product to be sure it does EXACTLY what you want, just like you always did. And, it’s still a good idea (and as possible as ever) to keep the product modularized and also check every module to ensure it works as you want. How small you make those modules has always been a matter of preference / dictated by levels of concern, etc.
If you are generating tests with an LLM, it’s not deterministic.
Also, have you actually read anything an LLM wrote that’s longer than a hundered words? It’s usually immediately obvious. In the open source world things I keep seeing are exchanges like this one. And Niri is a fairly good case with it’s maintainer actually reading the PRs and catching all the bullshit that an LLM spits out. Very often nowadays people will also do reviews with an LLM and this sort of stuff will just accumulate in the codebase with nobody to reject it.
But even if it was as useful as big tech paints it to be, I’ll always be against using it based purely on morals. There is an enormous number of issues in that regard, most importatnly LLMs literally kill people both directly and indirectly. Condoning the adoption of this technology is either being ignorant or misanthropic.
If you are generating tests with an LLM, it’s not deterministic.
Once the tests are generated, they have become deterministic - unit test code is checked in alongside the repository, CI (continuous integration) systems can prevent the code from merging to main branch until all CI tests pass. No LLM is involved in the test execution process, and if all LLMs stop functioning tomorrow, the tests still work just like they did before.
read anything an LLM wrote that’s longer than a hundered words? It’s usually immediately obvious
About 250,000 lines of code so far, and no… I don’t read it all - my personal music player isn’t that important, I just care that it works. Other more important projects I have read every single line. Even in the music player I have read a few hundred lines and skimmed a couple of thousand. Yes, the style is obvious. It’s what you might call “too complete” attempting to cover lots of edge cases that a human would usually neglect, though as human code is maintained over the long haul, it tends to mature in that direction. And just because it’s very complete doesn’t mean it’s perfectly correct, and that’s a major point of conflict with me and LLM code, I’m constantly asking it to simplify because the complexity is easy to hide mistakes in. You know what happens when you ask it to simplify? It simplifies.
people will also do reviews with an LLM and this sort of stuff will just accumulate in the codebase with nobody to reject it.
That’s a management / maturity problem. As new as all this stuff is, it’s virtually inevitable there will be an adjustment process. Our company runs on processes and procedures, they average 50 pages each and there are thousands of them. I am given almost 100 of them to “train on” on a near annual basis - whenever they get revised. No way in hell can I do an accurate review of 5000 pages of intertwined procedures, not code, natural language work instructions. LLMs can. LLMs are. Simple questions to an LLM like: “review this procedure, is it internally self consistent” will find issues, real issues - usually minor - but undeniable when you read the review, the problem it describes is absolutely present in the document, page 14 says “X” and page 23 says “not X” and the context is basically identical - so which are we really supposed to do: “X” or “not X”? Well, that depends. Depends on what? Additional context and detail that the already too long 50 page procedure doesn’t address. We just did an internal audit which turned up thousands of these mostly little issues. Addressing them is going to be… interesting. We’ve been in business over 70 years, we know we’re not going to tackle it all in a single year, there will probably be a plan to work down the worst of the issues, then a second round for additional polish, of course corrective actions to ensure that we at least don’t make things worse along these metrics with future changes… I bet by 2035 we’ll have 90%+ of today’s real observations fully addressed.
LLMs literally kill people both directly and indirectly
As does nearly every activity of life, both in human society and the natural world.
Why the are you getting more text than you can physically review at your job? That sounds like a recipe for poor reviews. How are you sure LLMs are not missing anything if nobody actually has the time to read through all the text? Why does the text that gets to a developer contain logical mistakes in the first place? Back at my last job we had people assigned to collecting feature requests, talk them through with whoever wanted the feature implemented and produce proper specifications that developers would implement without wasting time on specifications containing garbage.
As does nearly every activity of life, both in human society and the natural world.
Yea, all people who drank water died. Not a very good argument. Most things don’t have a mental disorder named after them. If I leave a webpage with a calculator running unattended it won’t magically persuade a random visitor into commiting suicide. Webpages with LLM-powered chatbots are doing just that. You may choose to neglect it, but to me it sounds like adding lead to your gasoline. Yea, your car runs better. Are you sure it’s worth it, though?
Why the are you getting more text than you can physically review at your job?
Regulatory required documented training, old conservative company.
How are you sure LLMs are not missing anything if nobody actually has the time to read through all the text?
You’re not. Neither do the humans tasked with the job do a 100% thorough review. Side by side, the LLM catches more flaws than the humans do.
produce proper specifications that developers would implement without wasting time on specifications containing garbage.
Sounds nice. We’ve got ~500 employees at our “site” (which is now distributed across 3+ states with remote workers), of those, about 8 do software development, so developers end up doing virtually all of the translation of customer “needs” to specifications / requirements. We also have a marketing department that changes their mind about twice as fast as the product development / delivery cycle.
If I leave a webpage with a calculator running unattended it won’t magically persuade a random visitor into commiting suicide.
That’s a touchy subject. A friend of ours recently had her son die of drowing in his bathtub, he was epileptic, had a seizure and fell in the tub. There was no suicide note, it was ruled an accident. He used ChatGPT to write his own obituary, as well as a book explaining his life from his perspective - each seizure was “taking away a little of who he was,” and the seizures were getting more frequent, none of the treatments or medications slowed them. It’s pretty clear, to me, that he was going to commit suicide and/or die soon with or without ChatGPT. After he passed, his mother had a ton of grief / guilt about not being closer with him in his life, but then… he actively pushed her away, even violently. It’s doubtful he would have written or expressed what he did as completely without ChatGPT assistance - how much was him, how much was LLM hallucination? Does it really matter? He wrote the book, and he put it where she would find it.
The next time you go driving, count the oncoming cars. When you get to 100, realize this: statistically, one - maybe two - of the drivers of those oncoming 100 cars will, eventually, end their life by suicide. And those are just the deaths officially classified as suicide. Engagement with ChatGPT undoubtedly works both ways, some cases accelerating the process or even causing them to take their lives when they wouldn’t have, and other cases slowing or stopping the pursuit of suicide altogether. It’s going to be hard to say if Chatbots or other world events drive the statistical changes starting a couple of years ago. The data isn’t clear yet, but the concerns have certainly been raised, and I would assume the supportiveness of Chatbots should improve as a result in the future.
But you really should be writing those tests regardless of whether or not you or an LLM ends up writing the actual code. That helps ensure that the code actually does what you think it does, and prevents bugs or unintended changes in behavior from sneaking in later when something changes
Well, depends. My thousand-line pet project does not need any tests, it needs to be feature complete and never touched again.
Bigger projects are usually ones where multiple people are contributing and having code that practically nobody wrote there is a huge problem on it’s own in my opinion, even if it passes a number of tests. Good tests themselves are usually harder to write than the code they are testing, I am not sure if delegating the actual fun part of development to an LLM and doing “chores” manually is a good idea.
Indeed, my code has never been more rich in unit tests now that coding assistants make them so easy to add. If the coding assistant subsequently does something weird that changes program behaviour it knows immediately.
Indeed, my code has never been more rich in unit tests now that coding assistants make them so easy to add
Me as well. A year ago there was significant trouble with those unit tests not adding much value but that has improved a lot over the past several model updates. The unit tests aren’t 100% as good as they could or even should be (neither are human generated unit tests that take 10x as long to add) - but in terms of time invested vs value derived, they’re excellent value for effort.
If the coding assistant subsequently does something weird that changes program behaviour it knows immediately.
About 90% of the time for me now, in Opus 5.5.
What do the tests do?
They check to make sure that the code continues producing the expected results. For example if I had some code that added up numbers, I might have a unit test that calls that function with the parameters “2” and “2” and checks to make sure the result is “4”. If something changes that makes it start returning something else then a flag is immediately raised.
Are these tests written by AI? Have you reviewed the tests? Do they cover all code paths and all the possible inputs and outputs? Do the tests have requirements cited (traceability)?
Unit tests have been around for a long time. They tended to be under-utilized in the past because they’re such a maintenance headache - changes / improvements in the production code that used to be a simple 5 minute thing can easily balloon to an hour or more of headache in necessary adjustments to perfectly good unit tests for “the old way” that happened to be impacted by the change.
Unit tests are a good way of keeping under-experienced programmers from accidentally breaking old things as they work on new things. They are VERY appropriate tools for LLM agent assisted development.
Absolutely not. In a world where hallucinating machines are writing code, humans should be writing the tests at a bare minimum.
AI writing code to pass tests written by AI will get people killed.
Why does everyone who starts using AI always devolve into posting like this?
it’s crack for burnt out programmers who don’t enjoy the process of creating software anymore. You get to have results without putting in any work of your own. In corpo-world where everything is a metric, outputting massive amounts of code quickly and effortlessly is a great thing. For the others who actually enjoy making something it’s a useless tool.
I generally try to avoid using AI for my hobby projects, but I think there is still a niche for it in working out issues that take a disproportionate amount of time to track down manually relative to the scale of the fix. Things like a subtle typo somewhere or missing some hyperspecific nuance of whatever stack I’m working with. In those particular cases I would say it actually enhances my own creative ability insofar as I don’t need to spend hours of effort banging my head against a wall until I eventually figure out what went wrong, and I can put that time and mental energy into the stuff that’s actually interesting to me.
But yes, in general I do agree it’s kind of useless otherwise for creative endeavors or passion projects. It’s great for helping with code being written solely for a paycheck, but for personal projects it largely defeats the purpose of working on them in the first place. I think the trouble is that in general AI does make a lot of things easier and I have a lot of anxiety about getting into the habit of overusing it in that way, so I try to keep it as a sort of last resort.
Maybe because when they actually try them out they’re discovering that AI coding agents are pretty good?
We’re all being hypnotized by aliens who have replaced our political leaders and oligarchs, and now they’re using LLM chat hypnosis to take over the next tiers of society. /s
Probably because they get pissed off by the illogical anti-AI crazies.
This might get me banned here but I mostly agree with him
AI is a tool and the problem world wide is that nearly everyone is using the tool wrong
But used right, it can be a huge benefit. We know how deeply checked and tested all code is (who doesn’t remember the continuous Linus rants?) so why does it matter if submitted code covers a bug that was discovered through AI, or even written by AI? As long as the code checkers know and understand what happens, I dont see the problem with the code having been written by a tool
Obligatory AI doesn’t exist, its LLM’s, and LLM’s are tools, and tools are fine as long as you use them well.
If companies suddenly started making two meter long chainsaws and people started running around with these chainsaws cutting little girls in half, then yeah, people would rightly question if we shouldn’t ban those tools too.
Force AI companies to be responsible for how their products are used and wait to see how fast things would get better.
And for developers: as long as every commit is reviewed and checked by humans, to me its just another tool being used to make Linux better
For those arguing the water of energy and the theft of intellectual property, it’s the same point: governments need to put rules in place to control the stupid growth of data centers, AI companies must be held yo account for the data they’ve stolen, absolutely. However, that doesn’t mean that LLM’s are tools without any use
He’s right about the tool aspect of it, but, and I cannot emphasise this enough, it is not just a tool. Though we all bow (or salivate) before our fake materialist gods, and cast aside the beauty and relationships in the world, things are more than just their function.
AI is a tool
If you mean “only a tool”, this is wrong. See for example https://www.youtube.com/watch?v=yZUuKzDQSsI (ignore the stupid title)
This might get me banned here but I mostly agree with him
I have seen a turn in the tide in the past few months. Whereas online forums discussing LLM tools for coding used to be near 100% brigade against them with mostly bluff and bluster and no real basis for the arguments, attitudes have been softening. I actually read one rather long thread with 90%+ support for the tools - not blind enthusiasm with no evidence, but clear statements from users about what is working well for them and what still needs improvement. All in all, work with the new tools seems to be going better “in the trenches” than work without used to.
In the end, it doesn’t matter if you wrote the code, an LLM wrote the code, your cat tapped it out, or aliens communicated it by blinking their home sun… you check it, it works to your satisfaction, you submit it for review and merge for use by others. In all those scenarios, it’s “your code” - the LLM, your cat, aliens, none of them are going to stand up and defend the quality of the product. If you don’t really understand why what your cat wrote works, you probably shouldn’t sumbit it. Especially the alien code should be carefully examined before running, but at the end analysis it’s the code that matters, not its provenance.
Generated code was trully bad. These days? Not so much anymore…
Better tools, more acceptance.
2024 I’d use “the tools” to generate snippets that I would copy-paste and massage into my larger projects. Those snippets tended to be 2 to 6 lines long, and they’d be usefully correct about 80% of the time. I laughed at people who were “vibe coding” with those tools, even in early 2025. Like Linux in 1994, I’d call it “interesting, not ready for prime-time.”
By mid 2025, I saw the potential for improvement - but the better the code output got, the louder and more irrational the online reactions against it got - up until the past month or so it was like a guaranteed brigade of negativity.
AI is a tool and the problem world wide is that nearly everyone is using the tool wrong
Maybe if the tool is so finicky that it’s practically impossible to get right, it’s a bad tool? For sure millions of programmers are just prompting it wrong… Can’t be that it’s just not as great a tool as the AI overlords claim
As long as the code checkers know and understand what happens, I dont see the problem with the code having been written by a tool
Until the code checkers resort to LLMs because the agentic coders submit so much code for review. I will just let AI run over this one time, what could go wrong? I’ll tell AI to generate test cases beforehand to make sure it gets the implementation right. Bad tests never lead to problems right?
Or maybe, just maybe people eating up a constant stream of propaganda produced by those whose whole financial future is tied to AI being the universal solution to every problem (and some they have to create in the first place) are indeed a problem.
Just because 24/7 advertising tells you to hammer in screws doesn’t mean it’s the smart thing to do. And neither does it show how useless a hammer is as a tool.
AI is a tool and the problem world wide is that nearly everyone is using the tool wrong
Because the public discussion devolved into some insane fight between people ideologically against any AI (funnily often the argument “it’s stealing other people’s work” is coming from people that are fundamentally against copyrights in every other case) and some not less crazy people believing that AI is their new lord and savior (and a lot of them are paid or the followers to said paid influencers you just changed one idol (tech billionaires) to another one as instructed…).
It’s basically a mud fight between those believing that hammers are inherently evil and need to be destroyed and those that think everything in this world is a nail. Nothing in-between.
It’s so funny when they try to argue that the conditions around it are somehow different and make it unethical. Like yeah, I’m against fascism and climate change, but all our previous tooling was also built on unethical grounds, this isn’t any different. You like trains? The railroads were built by slave labor, extracting the fuel displaces indigenous people and hurts the environment, and folks like Carnegie used their train-powered wealth for their own personal benefit to the detriment of the public. Does that mean taking a train is unethical?
Things were bad before AI, so let’s make them worse.
He’s drunk the AI Kool-Aid.
That’s a good thing, it was Flavor Aid that was laced with poison in Jonestown.
wow… his ego is pretty big and bursting. Good thing that he evacuated his brain power to LLMs.
Well, it was nice while it lasted.
*slaps an orange flag over the tux sticker on my laptop*
2027 will be the year of the NetBSD desktop, guys!
He’s been like this for a while now (at least a few months), I admit it affected my choice of OS for a new machine.
I use FreeBSD btw.
NetBSD hasn’t fallen into AI psychosis but it still has some slop in it.
Year of the GNU/Hurd desktop!NetBSD has one slop commit, and they have apparently since removed the code it brought in. Besides, I need some level of mainstream software support for uni, and pkgsrc is half-decent for that.
GNU Mach (a component of Hurd), meanwhile, have been using Claude for code reviews (https://lists.gnu.org/archive/html/bug-hurd/2026-03/msg00100.html, https://lists.gnu.org/archive/html/bug-hurd/2026-02/pdffAOvuZusjj.pdf); and the GNU Project itself is still on the fence (GCC, for example, has 57 slop commits and counting).
I’m switching to haiku.
I plan to put Haiku on my old ThinkPad; but the lack of even a lock screen in the OS sadly makes it unsuitable for my main machine.
For me the most criticism against AI is rather criticism against Big Tech. AI itself however doesn’t have to inherently come from Big Tech so banning it wouldn’t really be reasonable for something like the Linux Kernel.
It doesn’t inherently have to come from big tech, but currently there’s nowhere else to get it, soooo…
Maybe we ban it now and then revisit all this after small open source local models are actually useful for anything?
Not true. I’ve been coding for months now using only local models running on my own hardware. It’s perfectly adequate with the right harness.
Ok well that’s actually awesome!! But is that what these people are using in big important projects like the Linux kernel, or are they using Claude? I’m guessing it’s the latter, right?
Local models are capable of assisting coding in even enormous, complex pieces of software. That’s where having the right harness really shines - it provides the model with the tools it needs to focus in on just the bits of code it’s currently working on. Sort of like how a regular human programmer can do a much better job working on a large project with Visual Assist or equivalent in their IDE. I’ve also installed some tools like Context7 that provides the LLM with API documentation (no more hallucinated APIs) or Firecrawl (it can just go search whatever it needs like a regular person). It’s not so great at large-scale architecture questions, but that’s not what 90% of coding usually involves.
If you want a step up from what can run on just commodity desktops, some of the bigger models are also open-weight and a company or studio with a programming department could run one of those on their own hardware in the server room. GLM 5.3 is right up there next to Claude in the benchmarks, for example, and doesn’t take a data center to run.
Or you could go halfsies on the “big tech” thing and have an API subscription for a big model that your coding harness uses for doing the large-scale architecture planning and then hands off to small local models for implementing the plans the big model wrote. That’s a very common pattern too.
The key point is that there’s a lot of options here, it’s not “pay Anthropic or nothing.”
Of course! But also, AI shouldn’t be used to simply generate code. It should assist human developers into solving problems like figuring out why your code won’t build, or why certain errors occur or detect bugs.
Why? Because if everyone starts using AI to generate code, you lose experienced developers who will understand what’s happening in the first place. You lose the expertise. And that’s a big problem.
Yeah, and as I understand it that’s what Linus wants people to do. So banning AI would not allow this to happen when working on kernel code.
AI will come for Linus too someday.
LLMs aren’t new or interesting.
Linus being Linus :)
