- cross-posted to:
- linux@lemmy.ml
- cross-posted to:
- linux@lemmy.ml
Imagine taking the lightweight, privacy-focused appeal of Linux and completely ruining it by hardwiring AI models straight into the OS. Nobody is switching from Windows for more forced AI features.
You must log in or register to comment.
The appeal to Linux is you can do whatever the fuck you want with it.
That means some people will use it to do something stupid with, you can’t stop them and nor should you.
So like 7 PCs assembled by the same QA guy probably.
Tell that to the kernel’s
BDFL.what happend to B ?
-is it like marking over the letter with a line or something ? let me check
—B
whatever
Tech-bros gonna tech-bro. Go on Kickstarter and see all the random shit, “With AI!” It is fucking pathetic.
and then you see the 10000% funded… ffs, not everything needs AI …
Challenge windows? By also shooting yourself in the foot with the same gun?
tiktok style shoot your own foot challenge
Ai worked really well for windows…
It’s very memorable. Everyone has feelings about Copilot, especially that key that can’t be turned back into right control.
Isn’t that what Omarchy was already doing?
Another criticism levied against the project is that it does not meaningfully build upon its base Linux distribution, as it consists mostly of configuration for existing open source software.
From the wikipedia page. I didn’t see anything about AI, just a lot of donations from CEOs and some horrible opinions about immigration.
Forrest Knight made a video about it. From my understanding the appeal is that you can introduce new features, fix bugs and customize the operating system with the agent(s).
Call me what you want, but the idea of a Linux distribution where you can modify config files using natural language like “make my taskbar 25% bigger and map meta+E to a new yazi window in terminal” without having to rummage in /etc for config files or hunt a specific command syntax sounds like something that would help both the seasonned sysadmins and newbies make the most of their distributions.
Sure, a mistake could take you to grub rescue, but we could easily build guardrails around this, and I was already doing it to myself without the help of AI, so…
Yes but Omarchy fails so completely on every conceivable level that it doesn’t actually fill a market niche
That hasn’t stopped it getting more dollars per unit time since it started existing than any other FOSS project in history though
TBF, Omarchy is also run and funded by raving rightwingers. Until further notice, I’m not going to assume out of the gate that this project is too (but fascists do love slop).
So they hope RAM prices will come down enough to have people afford the machine needed for it on release…
Linus vibe codes nowadays, so it’s already in the kernel.
Maybe it’s time for Linus to retire
While that is true, we don’t have to accept more slop! There are other kernels that exist which actually follow better practices. We can have microkernels, we can and should reject more slop in Linux. I don’t want it to turn into Windows! I will learn how to make my own operating system if need be. ROFL I am learning Python and C proper, so the ability will come to me soon.
I will learn how to make my own operating system if need be. ROFL I am learning Python and C proper, so the ability will come to me soon.
Oh boy, new TempleOS incoming!
This is pretty cool
Think about where TempleOS would be if Terry was around in the golden age of vibe coding 😏😏 /s
I will learn how to make my own operating system if need be. ROFL I am learning Python and C proper, so the ability will come to me soon.
All the best
To me AI, is a tool like any other. The output turns out sloppy if the human wielding the tool it isn’t doing their job properly. Quality control is a responsibility that never went away.
First of all, no it’s not. A hammer is a tool. There’s only one person responsible for taking it and making it useful: you. The fire department is not a tool. It performs a useful function which you can initiate, but it requires a complex chain of logistics which involves large amounts of resources and specially trained humans at every step to continue functioning. AI is a service.
Second of all, even if we accept that it is a tool, a jackhammer that has a specialized tip made from a rare material mined almost exclusively using slave labor which needs regular replacement and which runs exclusively on a smelly gasoline engine which everyone around you then has to breathe is not a tool “like any other” and using it is not a neutral act. When you use Claude, you are giving money and support to a company that buys, scans, and then shreds physical copies of rare books; scrapes independent websites until their servers give out, actively circumvents webmasters’ attempts to block them, and then lies about respecting opt-out signals; builds datacenters in disadvantaged areas that do everything from raising power bills and making power service less reliable for nearby residents to raising the temperature of places up to a mile away and producing inaudible infrasound that causes everything from irritability through hearing loss to siezures; powers those datacenters in part using natural gas turbines that produce significantly less electricity and more pollution than the same amount of natural gas in a natural gas power plant; and sells their technology to militaries which use them to design guided bombs.
And that’s before we even get started on what every single study on the cognitive effects of prolonged LLM use, even by trained professionals automating things they can do themselves, has revealed (in a sentence, use that muscle or lose it).
Fair enough. In a broader sense, all of that is true. The current implementation and business practices are completely rotten to the core. That’s why I’m not going to pay any company for AI, unless they publish a report about using renewable energy. Mistral AI is pretty transparent about their emissions, and that’s a step in the right direction. Still not good enough until they actually power their data centres with renewable energy.
Anyway, I was mainly talking about LLMs specifically, not the entire AI industry and their destructive habits. None of the current AI companies deserve any of my money, especially the American ones.
First of all, not giving them money is like pirating a Harry Potter show and then posting about how much you liked it on social media. You’re still giving it free advertising to people who will spend money on it.
Second of all, see what I said about the cognitive effects. If you don’t practice doing things without the AI, you will lose the ability to do them much more quickly than you think you will, and if you start relying on it to make tiny decisions, at best your work and at worst your thinking will slowly start to reflect the model’s biases.
More importantly, programmers who use LLMs to complete programming tasks demonstrate less complete understanding of their code, which is important because eventually you need to maintain it and even LLMs’ most ardent defenders admit that all of them are dogshit at software engineering. If you use AI to maintain an AI codebase, you’ll do great for a couple of months before it loses the ability to remember all the parts and starts duplicating work and not syncing up state with other components and before long it’s a buggy unmaintainable mess.
And just like a tool, it can be too dull, too sharp, or just wrong for the job.
Oh, absolutely. The person wielding it needs to know what they’re doing. You don’t hammer nails with a screwdriver unless you’re a total masochist. IMO current LLMs are ideal for generating the filler for self-help books. You have like 10 pages worth of ideas to say, but the publisher demands a 250-page book.
IMO current LLMs are ideal for generating the filler for self-help books. You have like 10 pages worth of ideas to say, but the publisher demands a 250-page book.
lol okay. For a second there, with that previous comment you made, you made me think you actually thought LLMs were good for coding.
Well, it depends. Mostly, I’m against it, but there are ways to vibe code responsibly. My requirements are very high. I don’t think most people vibe that way, hence the slop code you see out in the wild. If you’re using an LLM to write your code, I expect you to read and understand every line thoroughly. The programmer should be prepared to answer tough questions about any part of the code as if they wrote all of it.
If you actually take the time to do that, you’ll either request a few more iterations of updates or fix it yourself. In reality though, people can’t be bothered to do it properly, and in that context vibe coding would be banned.
THANK YOU META
OK i use local models so… unless it has loads of really small models for single tasks (that eat up disk space) or it has some all purpose ones that will eat all your GPU and Ram OR it APIs data centres costing you a fortune…
Also giving AI admin…yeah good luck with that
It’s not quite good enough for any real practical purpose IMO but I would be curious to see what sort of experience something like the liquid AI LFM3 concept extended to a whole ass operating system is like.
