Surrendering generative AI and all of its utility to the ruling class because of its disruptive effect on intellectual/artistic labour is a folly, it’s luddite tactics that we know come from a good place but are ineffective and leave its advocates in the dustbin.
We are marxists and as such the material world and its realities are most important to us. The fact AI is capable of spitting out shortform content more efficiently than ever before cannot be neglected or the ruling class gets another cudgel under its belt. Why not produce propaganda using the same avenues as the elite? In addition, letting something like ‘fash aesthetics’ influence you is absurd. If I can eek one positive thing for the working class out of ‘fash aesthetics’ then I really don’t care about said aesthetics.
Another thing - the ‘cudgel’ under the belt of the ruling class - at the moment generative AI is dominated by paid services and companies controlling its input and output. Despite this, all forms of generative AI can be handled locally. AI running on your own machine truly is just another tool at your disposal. If we don’t utilize and educate about generative AI then the ruling class keeps its grip and knowledge of the ‘true democratization of art’ and any benefit is lost.
@MasterBlaster I didn’t mention its disruptive effect on intellectual/artistic labour. I brought up the fact that these generative platforms are created and run by companies who have their fingers wrist-deep in current successful far-right movements, and that the AI aesthetic has become the aesthetic of these movements.
I don’t think Yogthis is using a generative LLM produced by China on their computer. I looked around; there are two of them, and neither seem to have their source published on an English-speaking website. So unless one goes through the trouble of tracking down working links for ERNIE or Alibaba’s generative AI source codes and implementing them off of Chinese instructions, you are using the software and the aesthetic of these fascist companies, therefore normalizing it and benefiting it.
Edit: God, they even responded to someone asking for the source with an AI-generated answer that only links chatbot source codes and only touches on one of the above two mentioned generative AIs out of China (which, again, doesn’t have a working link that I could find; it’s based on a model by PaddlePaddle, which you can get, but ERNIE’s text-to-image AI itself seems to be gone from the English internet).
I don’t think Yogthis is using a generative LLM produced by China on their computer.
Yes, that’s exactly what I’m doing. I run DeepSeek and Qwen models using Ollama locally. They work great. I also use full DeepSeek online. It’s absolutely bizarre that you would make this assumption without even asking. I also run Stable Diffusion models locally using https://github.com/AUTOMATIC1111/stable-diffusion-webui
I looked around; there are two of them, and neither seem to have their source published on an English-speaking website.
No you haven’t because if you did then you’d quickly find plenty of Chinese models ready to use that are in English. I’ve linked a few in this comment https://lemmygrad.ml/post/8454753/6668311
You’ve literally done zero investigation before spewing nonsense here. It’s incredible to see such low effort trolling on here.
@yogthos Ollama is, if I recall, an American project sponsored by Meta. And Stable Diffusion, as already mentioned, is American too. So, after your very first post to me in here was proclaiming that you’re using Chinese software, now you admit that you aren’t.
In your post there, you have two text-to-image generation models. Again, one of them I’d found (and is the one that I could find nothing but broken links for), one of them I hadn’t.
(Edit: Two that I had; I see Qwen is the one that Alibaba is released, so, my bad on reading comprehension)
I’m not lacking in research, you’re lacking in honesty. And you accuse me of low effort trolling when you’re literally using AI to manufacture replies to me? Dude.
Ollama is an open source project. The fact that this tech originates in the US does not mean it shouldn’t be used. Most of the software and hardware you use is of US origin in one way or another. The discussion was about the models themselves, which are of course Chinese. You are shamefully ignorant on the subject you’re attempting to discuss here.
The only one lacking in honesty here is you bud, and you ain’t fooling anyone here.
On another note Ollama is really shitty and they’re trying to push their own API format, please consider using the one true community deriven gigachad Russian developed ‘llama.cpp’. KoboldCPP even has a nice web UI for use.
Surrendering generative AI and all of its utility to the ruling class because of its disruptive effect on intellectual/artistic labour is a folly, it’s luddite tactics that we know come from a good place but are ineffective and leave its advocates in the dustbin.
We are marxists and as such the material world and its realities are most important to us. The fact AI is capable of spitting out shortform content more efficiently than ever before cannot be neglected or the ruling class gets another cudgel under its belt. Why not produce propaganda using the same avenues as the elite? In addition, letting something like ‘fash aesthetics’ influence you is absurd. If I can eek one positive thing for the working class out of ‘fash aesthetics’ then I really don’t care about said aesthetics.
Another thing - the ‘cudgel’ under the belt of the ruling class - at the moment generative AI is dominated by paid services and companies controlling its input and output. Despite this, all forms of generative AI can be handled locally. AI running on your own machine truly is just another tool at your disposal. If we don’t utilize and educate about generative AI then the ruling class keeps its grip and knowledge of the ‘true democratization of art’ and any benefit is lost.
@MasterBlaster I didn’t mention its disruptive effect on intellectual/artistic labour. I brought up the fact that these generative platforms are created and run by companies who have their fingers wrist-deep in current successful far-right movements, and that the AI aesthetic has become the aesthetic of these movements.
I don’t think Yogthis is using a generative LLM produced by China on their computer. I looked around; there are two of them, and neither seem to have their source published on an English-speaking website. So unless one goes through the trouble of tracking down working links for ERNIE or Alibaba’s generative AI source codes and implementing them off of Chinese instructions, you are using the software and the aesthetic of these fascist companies, therefore normalizing it and benefiting it.
Edit: God, they even responded to someone asking for the source with an AI-generated answer that only links chatbot source codes and only touches on one of the above two mentioned generative AIs out of China (which, again, doesn’t have a working link that I could find; it’s based on a model by PaddlePaddle, which you can get, but ERNIE’s text-to-image AI itself seems to be gone from the English internet).
Yes, that’s exactly what I’m doing. I run DeepSeek and Qwen models using Ollama locally. They work great. I also use full DeepSeek online. It’s absolutely bizarre that you would make this assumption without even asking. I also run Stable Diffusion models locally using https://github.com/AUTOMATIC1111/stable-diffusion-webui
No you haven’t because if you did then you’d quickly find plenty of Chinese models ready to use that are in English. I’ve linked a few in this comment https://lemmygrad.ml/post/8454753/6668311
You’ve literally done zero investigation before spewing nonsense here. It’s incredible to see such low effort trolling on here.
@yogthos Ollama is, if I recall, an American project sponsored by Meta. And Stable Diffusion, as already mentioned, is American too. So, after your very first post to me in here was proclaiming that you’re using Chinese software, now you admit that you aren’t.
In your post there, you have two text-to-image generation models. Again, one of them I’d found (and is the one that I could find nothing but broken links for), one of them I hadn’t.
(Edit: Two that I had; I see Qwen is the one that Alibaba is released, so, my bad on reading comprehension)
I’m not lacking in research, you’re lacking in honesty. And you accuse me of low effort trolling when you’re literally using AI to manufacture replies to me? Dude.
you’re doing an awful lot of talking out of your ass in this thread
@PoY Whatever. Maybe I am, if I’m the only one here who thinks it’s obvious that none of this adds up. I’ll bow out.
Ollama is an open source project. The fact that this tech originates in the US does not mean it shouldn’t be used. Most of the software and hardware you use is of US origin in one way or another. The discussion was about the models themselves, which are of course Chinese. You are shamefully ignorant on the subject you’re attempting to discuss here.
The only one lacking in honesty here is you bud, and you ain’t fooling anyone here.
On another note Ollama is really shitty and they’re trying to push their own API format, please consider using the one true community deriven gigachad Russian developed ‘llama.cpp’. KoboldCPP even has a nice web UI for use.
oh thanks for the heads up, I’ve heard of llama.cpp, but never bothered looking into taking it for a spin.