- cross-posted to:
- politicalmemes@lemmy.world
- cross-posted to:
- politicalmemes@lemmy.world
cross-posted from: https://piefed.zip/c/politicalmemes@lemmy.world/p/1758477/using-fascist-technology-to-fight-fascism-still-promotes-fascist-technology
Using fascist technology to fight fascism still promotes fascist technology.
Brother you can run that shit local and all of your claims have so many asterisks attached that they’re essentially pointless. Everything you claimed is entirely true of base computers no ai. Hell even just internet access has been shown to have all the same negative effects. And I’m defending it because I’m not stupid and it’s a useful technology. Run your shit local. The tech bro goons don’t own the technology.
Ok, you have a local model… but did you train it yourself? On what data?
Sadly, modern generative AI is a tool that’s built on top of the exploitation of other people’s work.
Actual human beings put blood, sweat and tears into their words, art and music, only for it to be used without consent or compensation by unskilled lazy misanthropes to generate an infinite sea of unwanted slop.
There is theoretically an ethical way to do AI (using public domain, creative commons, licensed or original data), but it seems to me that almost nobody is doing that.
Personally I don’t care how “useful” people want to claim AI is. For one, I’m not seeing things getting better faster, if anything they’re all getting worse. And regardless, the ends don’t justify the means, so if we’re exploitint people to generate slop I don’t really care how nice the slop is.
Slop is a consequence of tool use, not a design feature of the tool. There are vast amounts of ethical data to train against. Etc, etc, etc. I could go on but you don’t really care, you care selectively. Or you wouldn’t be using the computer (the tool in front of your hands) undeniably built upon exploitation and compromise. It is a tool. It exists. You can’t stuff the genie back in the box. Work to make shit better instead of needlessly complaining about shit you can’t control like it’s some sort of generalized target you can paint on every iteration.
To me “slop” is a description that reflects the supply and demand imbalance of all generative AI output: there’s a practically infinite supply of AI generated stuff (to put it kindly), but very little demand.
For example, someone could generate a book with an LLM faster than any person could actually read it (assuming anyone actually wanted to read it in the first place…), and that represents a fundamental logical problem with regard to the value of things that AI produces.
Of course there are. Like I said in the comment above: works in the public domain, works licensed cc-0, original works, works licensed for training, and so on. Ethical AI is possible.
But is that what your local AI is trained on?
If so, cool and good for you, you’re using the tool ethically.
But that’s not what I’m seeing from >99.99% of AI users out there generating Ghibli-core slop for an audience of 0. Stealing all of Kazuo Oga’s background art is not curing cancer or doing anything good for society, it’s just blatant exploitation of one man’s life’s work and legacy so frauds and billionaires can make a quick buck off of people who can’t tell the difference between AI slop and human art.
This comparison isn’t exactly working in your favor…
Even assuming some of my PC parts were made in a sweatshop in Asia, how many people were exploited to make it? How badly were they exploited? Were they payed something for their time and effort?
Odds are that the model you’re using was trained off a mountain of work from of untold numbers of people, all over the world, who didn’t see a red cent for their skills, time or effort. Not only is their work being exploited for other people’s profit, nobody even bothered to seek consent to do it in the first place.
This is exploitation of labor on an industrial scale. AI turns the whole world into a sweatshop for every person in the creative working class. Forget talking about the “means of production”, creative types today aren’t even able to control the fruits of their labor anymore.
Like every technology, it can and should be regulated to prevent abuse, exploitation and other forms of negative impact. Every new technology introduces new problems, challenges and implications. But unlike what happened with oil and nuclear technology in the 19th and 20th centuries, we shouldn’t wait 50-100 years to consider them.
The big problem I have with most AI stans is that you are all so infatuated with the subjective utility of the “tool” that you’re unwilling to even have a good faith conversation about the objective impact of this technology on the world and society.
You just conceded my entire point. Ethical AI exists, it can be regulated, and the problem is how a tool is built and used, not the mere existence of the tool. The rest is meaningless boogeymaning. Your “99.99%” figure came directly out of your ass, and abundant supply does not make something worthless. Additionally, nobody said Ghibli knockoffs cure cancer. You picked the dumbest use case you could find and are pretending it defines the whole technology. You also turned “your computer is built on exploitation too” into “well, my exploitation is probably smaller and somebody may have been paid,” which is not remotely moral. I think it’s extremely smart to do some of what you mentioned though: regulate training, require transparency, license protected work. Great. None of that supports “do not use AI" or invalidates it as a tool. Oh and ending a comment full of invented statistics, cherry-picked examples, and arguments nobody made by accusing everyone else of bad faith is honestly… 🤌
They do own ChatGPT and other commercial LLM services, though, which is what most AI users use, and they made the hardware to run local AI models prohibitively expensive for a lot of people.
Cool so don’t use chatgpt and tools have costs? Idk what you want? Free compute and a lolly? Like I don’t disagree that those tech bros fucking suck and I personally don’t use their service but I’m not going to turn an entire technology into a boogeyman just because some fuckwhits with too much money and not enough empathy are fucking about. The underlying technology is solid and useful as a tool, regardless of the hardware you have. You can run local models that produce useful results on even a raspberry pi. It does not matter who contributed to it to assess it as such.
You cannot run it locally and have the same thing that oligarchs and big-tech have at their disposal.
You literally can. Kimi k3 is fist for fist punching at both sol and fable levels. The smaller qwen 3.8 27b is benching well against gpt 5.4 and opus 4.7. There are unreleased internal models you cannot run locally but you also can’t buy them. Additionally, big models are generally bad tools, you’re much better off tuning a local model for your specific task (Wendel from level 1 has a good video on this). If you’re gonna make claims at least do a Google search.
If you want something useable you need a decent amount of RAM/VRAM and a modern GPU, things that aren’t exactly accessible these days. That Kimi model is apparently half a terabyte in size and requires 100s of GB of VRAM, so I don’t know why you’re saying it’s a ‘local’ model.
That is patently untrue. You can have useful ai on a esp32, a $10 chip. https://github.com/slvDev/esp32-ai It will not be as capable as kimi k3 in general generic tasks but you can fine tune models for certain workflows (literally how “distillation” works) and you can run them on affordable and available hardware. Just because you don’t understand a tool or how to properly use it does not mean someone else doesn’t.
You brought up the Kimi model up in a conversation about local models being viable, something that requires 8 Nvidia GB300s as a prerequisite. And I’m sorry I don’t follow you’re little hobby close enough for you, I did try it and it sucked.
Kimi k3 can be ran at a cost of $37,900 with Mac studios. It’s not a hobby. Cool, glad for you. Tools are tools, I bet you don’t own or like an absolute gob of them but it doesn’t change their usefulness, and you don’t need kimi k3 for everything, I mentioned it specifically because it’s comparable to generalized frontier ai models and can be ran locally, but I also stated that you’re better off tuning models for your workloads specifically. Jfc why even comment if you’re gonna be so dumb and selective with your reading?
I was commenting based on my experience with this stuff, it was a lot of work to set up and the results I got were bad. I never even contended that it could be useful (but I’m reading selectively), I said you that you need decent hardware for it to be useable. I have an RX 6800 XT and 32GB of RAM and it was really slow, even with recommended models.
I tried to engage in good faith, and you proceeded to call me an idiot, sorry I reacted poorly to you being rude. You get what you put in.
Nah brother I’m not even going to continue giving you my time. You called my science and field a hobby, you questioned my reasoning when it was already presented, and you did not approach in good faith, you approached with preconceived notions, prejudice, and shit opinions based on your bad configs. I don’t owe you shit.
Of course it’s not the same. It’s better because I control its inputs and outputs, and who has access to them.
Oligarch controlled tools are dangerous because of who controls them, not because the tools are intrinsically dangerous.
That’s true, but they’re dumbasses. I don’t need Grok, or whatever awful shit facebook made, if I can localhost Qwen
There’s also Colibri which runs large models off of a SSD at a much slower speed than corporations. The large high-quality stuff has been gated by big-tech with gigantic GPUs. Colibri, if nothing else than a proof of concept, shows that high quality stuff can run on anything with a hit to speed.
No they don’t, but say whatever you need to ignore me. That’s definitely what someone arguing in good faith would do.
Citation needed.
Quite the opposite, actually. It causes enormous problems and has no benefits that aren’t better served by existing technologies that lack those problems.
Citation needed*
Well I’ll be the better man I guess… Okkkkkaaay so, first off, I’ve included a few studies published before the rise of modern ai, one from 2011 making extremely similar claims towards Google searches, but hey I can find more if you really want. Additionally, I’d love to see you work to replace several modern llm integrations with “existing technologies” because I’m here to tell you it cannot fucking be done and I’m willing to cite several fields but why don’t you just take a look at mathematics where ai has advanced the field several times over the past few months, or biology where ai has gotten so good at predicting protein fold structures that work for certain requirements that there is literally no alternative technology capable of the same feats.
https://pmc.ncbi.nlm.nih.gov/articles/PMC6866412/
https://pmc.ncbi.nlm.nih.gov/articles/PMC7099936/
https://onlinelibrary.wiley.com/doi/10.1002/wps.20617
https://scholar.harvard.edu/sites/scholar.harvard.edu/files/dwegner/files/sparrow_et_al._2011.pdf
https://aiinstitute.hbs.edu/navigating-the-jagged-technological-frontier/
The protein fold prediction stuff alone is huge. If we discover a cure for Alzheimer’s in my lifetime, or hell, even an effective treatment of the dementia symptoms, it will probably be due in part to these tools.
Well Google disbanded the AlphaFold team last month and moved the people there to work on Gemini to generate more slop, because they never really cared about that. It was just a stepping stone like making AI play Go.
Even if they reassign the researchers to other projects, I find it hard to believe that they just casually won a chemistry Nobel on their way to making more believable fake internet pictures. And putting aside the curing cancer of it all, why wouldn’t the most conniving capitalist in the world be interested in discovering patentable drugs? It’s huge business.