And just like that
G*mers are back to hating AI
That’s the problem when you don’t own
That’s a pretty significant part of the value and utility of AI…I bet stealing ppls ideas and getting to market first is still totally fine though, right?
Boycott Anthropic.
Gee. I wonder why anyone would want the inefficiency of local inference?
They’re training their models on my art. It’s only fair I get to use their technology in return.
Please, please make it useless for as many use cases as possible. Then maybe it’ll go away.
Was bound to happen eventually. They’re going to try to make open models illegal now.
My opinions are that its fine to make llm projects that threaten corporations who rely on saas. Doesn’t mean im going to use them or that they’re any good, but I want to see these companies who are firing all their employees and throwing all their money at ai to get scourned. Will build more distrust between the corpos too.
And just to be clear its okay when they steal literally the entire world’s knowledge base. But as soon as adobe and Microsoft see a 2% drop in profit all the sudden everyone’s a terrorist, or its the worst thing to happen ever. I hate ai but I’ll root for my enemy when he’s destroying one of my original enemies, can’t wait for these companies to burn. I’d go back to the stone ages if I knew elon, Sam Altmann, etc where all dead.
I’d go back to the stone ages if I knew elon, Sam Altmann, etc where all dead.
Going back to 80’s tech where most things were still analog should do it. Pen and paper, fax machines, analog tape and vynil records
I think we can realistically live in a happy medium with modern goodies. I do like apsects of analog and prefer it in many applications over digital. And its hard to say that in general because the first place my mind goes to is sensors for pi/esps, and your first thought is older analog tech. Analog sensors just seem more straight forward and it feels better to know you can just do shit without a computer in the way sometimes.
Like we sent people to the moon on kilobites of data, yet Microsoft windows needs 16 gigs of ram as a min req so copiolt can take screen shots every 30 seconds. Analog components just work on their own, require less power, and can just become its own system without a whole digital computer. Maybe instead of wifi bases lights and a whole home assistant server you just get a old fassion photoresistor and some clap sensors.
Its really as simple as just buying things that don’t need access to the internet or some dumb app just because the product costs less. I think in theory we could exist in a world where there’s a better happy medium where every devices doesn’t have to also be a android smart phone so that it can also spy on you and sell your data to ai and government.
But also at the same time I am stubborn enough where I will drag us down into the dark ages if it meant if the offer magically was presented to me. If you put two buttons infront of me I would press the dark ages button so fast.
Filofax, yes or no?
The best part of paying for a coding assistant is discovering halfway through your project that it has developed opinions about what your project is allowed to be. Imagine your IDE refusing to compile because it thinks your game feels a little too much like Tetris. At least a compiler doesn’t pretend it’s doing you a favor.
I think that’s what they wanted from the very beginning, to accelerate projects that are socially convenient for them, but not those that are not.
There’s that issue with universal machines that some people don’t like that others are allowed to make universal decisions.
I got shut down by it two days ago. It suddenly told me that the project I was working on had tripped some kind of guard, that it couldn’t continue working with me on it, and it couldn’t tell me why. I just went back to my local Qwen model. I can’t see this doing great things for selling Claude subscriptions.
As it was before AI. Local is the answer when shared is at the mercy of interests antagonistic to the users.
My coworker recently had an experience where he was trying to fix a security hole and his AI wouldn’t do it because it claimed that it was unsafe.
i know we’ve had our fun putting middle fingers up towards Adobe and the like, but these kinds of restrictions are needed to protect developers who aren’t insanely wealthy.
from the article, which i’m sure you all read:
And yet there are situations that merit some discussion, like the recent controversy about the Steam Next Fest being chock-full of AI slop copied from existing titles. That particular situation is getting particularly nasty, as game developers are reportedly being extorted under threat that their game demos will be cloned into complete games before they have a chance to finish and release the original.
Adding insult to injury, this is seemingly being done to small studios whose main priority is usually “having food,” rather than big publishers looking to microtransaction everything into a new yacht for shareholders.
now try to do the same and tell your boss you won’t do something because you don’t wanna
Local LLMs are the way to go on AI, they can’t suddenly change the model up on you. And also, I guess Anthropic just wants their models to be more useless ahead of going public? Because I presume people that want to use AI, who don’t care where they send their data to, but don’t want to nagged about how they can’t do benign programming and analysis tasks, will just choose another model provider.
They are the way to go, but hardware pricing makes it cost-prohibitive to match frontier capabilities, which makes decomp and recomp difficult. Or rather, they require more skill and take longer, introducing some barriers against the current floodgates being flung open.
We all know the eye-watering prices GPUs and RAM cost now, to say nothing of Unified Memory devices like the Strix. Most people just don’t have a spare $10K to throw at an inference box that can run (say) a 300B model at interactive (not batch) speeds.
Coming at this from another angle tho, I note that according to Steam stats, the most popular GPUs are in the 8GB and 16GB range…which is smack bang in the VRAM range of what a good MoE (say, Qwen 3.6 35B-A3B) runs on (for 8GB) or arguably the best local dense coder (Qwen 3.8 27B) can live at. Which leaves us with a capable assistant and a capable coder.
I think there’s probably room for LLM improvement in the 8–16GB range too…and given that most consumer PCs and laptops live there, I’m sanguine the open-weight devs might make it happen.
But I also suspect the gatekeeping of high-end capacities is intentional.
We’re probably not going to have the ability to vibe code Mario into Elden Ring again…not without significantly more skill. Dunno.
Uhm yeah good luck with running anything locally after hardware prices double again and again. Big tech needs to be stopped. There is no wiggling out of this just by being a smart customer.
Thing is, I already am…
The price did grow several times in size, but even in the 4 to 5 digit price range, you or smaller companies can have a pretty competitive performing open source model as the frontier models from earlier this year and they are keeping pace.
Obviously I’d like the speculation bubble to end and the DRAM and GPU cartels broken up, as it’s late-stage capitalism where the existing regulations being unenforced on big tech because the regulators are in on it that’s the problem, not so much the AI tech itself. (Don’t get me wrong, the tech has bad sides to it too like stealing artwork, data centre environmental impact and whatnot, but just speaking to the cause of the component greedflation).
Yes it’s disgustingly expensive. I spent a whole lot of money on an old used GPU and it felt bad, but it didn’t feel dirty like paying for Claude would. And then a couple of weeks later the used prices for this GPU had risen by about 15%.
(I only use Claude at work when someone else is paying. And they’re about to make me redundant because Claude is cheaper than me.)
Is there a local LLM I can run on an 8-10th gen i5 with 16GB RAM?
Yes, quite a few, but it depends on what you’re trying to do with them. Any ideas?
With an iGPU results are going to be very slow. Most promising things I have heard of so far is Mixture of Experts type models, if you have an 8GB card you might be able to run Qwen 3.8-35B-A3B-MoE which I think is the most capable for that hardware.
There are websites that can suggest a few for your specific hardware, I assume some are vibecoded and are only occasionally updated to the latest models, so take it as you will. See this Lemming’s comment. And !localLLaMa@sh.itjust.works is where I think most of the discussion around new development with Local LLMs happens here on Lemmy.
They suck far too much at the moment.
“Why have you done nothing for the past week?”
“Sorry boss, the AI refuses to do my work for me”
This is what happens when you rely on the products of giant corporations.
AI (restrictions) truly were inevitable.
But it’s not inevitable that the restrictions should say killing people is OK but analyzing how software works is forbidden.
Well that was an impressive speed run to enshitifcation. What kind of sensorship is queued up next?
This comment has been flagged for reason
[]. You may revert this conversation to an earlier point or switch to a non-american model
