- cross-posted to:
- technology@beehaw.org
- technology@lemmit.online
- cross-posted to:
- technology@beehaw.org
- technology@lemmit.online
You must log in or # to comment.
Am I the only one that read it for Hamas, by Hamas? Had to read it twice.
"said that the AI-driven traffic apocalypse is a nightmare for people who make content online…“If we don’t figure out how to fix this, the internet is going to die,” he said.
I make content online. I host my own web pages, I create my own content.
The difference is I don’t do ads and I don’t ask for subscriptions. Like the way it used to be. The internet won’t die, the internet for profit will. I couldn’t care less if AI scrapes my site, it has zero bearing on what I am doing.
The bigger issue, and the article is touching on it “journalism will die”. That is significant. Because any moron can post nonsense on the internet and have it picked up (also why the internet won’t die). Anyone with money in a different business can pretend to be publishing “news”.
I think they are right that AI is not going to save media companies, I don’t know who is going to pay journalists, and there are few laws about owning a monopoly on media companies and requiring them to tell the truth.
When people ask me what I want for a gift (birthday, Xmas) my only response now is “a gift subscription to [independent journalist-run media outlet]”. I figure if I’m paying for a couple of the ones I like on a rotating basis, that I can go grab archived webpages of the rest.
But it only works if enough people take up that social contract.
AI is a tool (sorry)
This should be a bumper sticker. Also, thanks for this, bookmarking 404, wish I had the means to subscribe.
My hope is that the “AI” craze culminates in a race to the bottom where we end up in a less terrible state: local models on people’s phones, reaching out to reputable websites for queries and redirection.
And this would be way better for places like 404, as they’d have to grab traffic individually and redirect users there.
My hope is that the “AI” craze culminates in a race to the bottom where we end up in a less terrible state: local models on people’s phones, reaching out to reputable websites for queries and redirection.
We’re already heading there. But, it’s not going to happen by sitting on your hands and waiting for the billionaires to hand you these local models on a silver platter. You honestly believe the overlords that own your phone will give you shit for free? They want you hooked on subscriptions, that send all of your personal data and social security numbers to their huge databases, until the day you die. And then they’ll sell that data to your children and your grandchildren just to make even more profit.
You have to take it. You have to find it yourself. You found Lemmy. Good. So, go find other shit. Discover open source. Discover piracy. Discover Linux. Stay on top of it.
Google just killed uBlock Origin, but I’m using Firefox, because the writing was on the wall at least a year ago.
I mean, I run Nemotron and Qwen variants every day, in a couple of UIs, and am experimenting with Hunyan and Falcon. You are preaching to the choir here :P
The NPU backends are not easy to write though. They really need extensive support from the phone chip makers, but fortunately, the HW makers are very interested in this exact future (and, specifically, selling SoCs for people to use it with).
Have you used any good ComfyUI workflows specifically for chat LLMs?
Not specifically. Ultimately, ComfyUI would build prompts/API calls, which I tend to do in scripts.
I tend to use Mikupad or Open Web UI for more general testing.
There are some neat tools with ‘lower level’ integration into LLM engines, like SGlang (which leverages caching and constrained decoding) to do things one can’t do over standard APIs: https://docs.sglang.ai/frontend/frontend.html
ComfyUI is just a bunch of Python code tied into I/O nodes. I’m surprised there isn’t a good set of nodes for SGLang yet.
SGLang is partially a scripting language for prompt building leveraging its caching/logprobs output, for doing stuff like filling in fields or branching choices, so it’s probably best done in that. It also requires pretty beefy hardware for the model size (as opposed to backends like exllama or llama.cpp that focus more on tight quantization and unbatched performance), so I suppose theres not a lot of interest from more local tinkerers?
It would be cool, I guess, but ComfyUI does feel more geared for diffusion. Image/video generation is more multimodel and benefits from dynamically loading/unloading/swapping all sorts of little submodels, loras and masks, applying them, piping them into each other and such.
LLM running is more monolithic: you have the 1 big model, maybe a text embeddings model as part of the same server, and everything else is just processing strings to build the prompts which one does linearly with python or whatever. Stuff like CFG and purpose-build loras do exist, but aren’t used much.
It’s a shame, because ComfyUI can be so much more than just image generation. And just because there’s a lot of string processing for LLMs doesn’t mean that it isn’t important to capture in an I/O interface, especially when it comes to preserving chat history. Save data, load data, ask new questions, etc.
ChatGPT is pretty damn powerful, I’ll admit. But, all of its components need to be localized, especially since something like a Mixture of Experts model could be split down to base models and loaded/unloaded as necessary.
If you get the paywall, the article is kind enough to let you know about removepaywall.com
