"We did it, Patrick! We made a technological breakthrough!"
A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.
AI, in this case, refers to LLMs, GPT technology, and anything listed as "AI" meant to increase market valuations.
The Western obsession with creating a "general" artificial intelligence will be its own undoing. It will stop itself when it eventually is forced to recon with the material realities of building infinite data centers and scraping infinite data from infinite monkeys on infinite typewriters.
Oh, it's far worse than that. They're going whole-hog on AI here too. Only...
...it's on a strict "prove your project or it gets cancelled" basis. There's none of this "let's just cram this down everybody's throat and see what sticks". AI projects, to get funded and permitted at all, have to establish what their goals are and what the failure criterion are. Without these there's no funding. With them, when, not if, the AI fucks up, the funding is cut and the project cancelled.
There's no doubling down on failure by pushing MOAR FAILURE!
Have you ever heard of the bitter lesson? Your suggestion has repeatedly been shown not to work. People want it to work so bad but the data doesn't bear it out: generalist systems beat specialists time and time again.
This doesn't make ai good for society, but specialist systems isn't the right path if what you want is things to be able to do hard tasks
You're just making shit up? A tool that does one thing well is always better than trying to make an omnitool that does everything.
If you need a hammer, you don't need a hammer that can also be a screw driver, knife, spatula, hair brush, and plunger.
See my other comment, but yes, this is a surprising result. It goes against many intuitions, but it's an extremely famous trend in the field.
generalist systems beat specialists time and time again.
Citations needed.
This is literally one of the most famous essays in AI (it has it's own Wikipedia article) and I mentioned it by name but sure here you go: https://www.cs.utexas.edu/~eunsol/courses/data/bitter_lesson.pdf
As for more recent stuff, people are doing experiments on it all the time, here's Nvidia very recently trying to figure out the best mix of training data (how much task-specific training, how much general-knowledge training for optimal results) https://arxiv.org/html/2606.24747
Google's 2022 "A generalist agent" - 1091 citations https://arxiv.org/abs/2205.06175
The entire field is built on statistics. It has many flaws, but this is like, the entire thing people are working on actively. Their goals may not be compatible with the flourishing of humanity, but finding the best way to automate various tasks is their one goal.
Even from gpt-1 people have been trying to make fine-tuned models for specific tasks and they keep failing compared to general models.
Elements of this idea do live on though, in MoE architectures, where they take a base model with knowledge of everything, then fine tune various versions of it for different things, and route your request to one of the models fine tuned for your task. This is mainly a workaround for the fact a large model with all parameters doesn't fit in memory so easily even in the massive Nvidia datacenter gpus, if it did, we can be pretty sure it would beat the smaller "experts" in most of the tasks
Also like, china isn't doing different to this? Deepseek (China) and glm5.2 (China) and mistral (France) and various other models are doing the same thing, because that's the thing that works (for the narrow definition of ai success that tech companies and politicians believe in)
All websites and services I managed are filled with poisoned data, text or images. At least 15% of my total processing power is spent on generating the poisoned text thata, but I'm glad to do it. Take more action people, not only "protest"
Tell me more about the poisoned text? I keep a site for a freelance job and I would love to fill it with poison.
This is the easiest one https://rnsaffn.com/poison3/ but there are more advanced ones that you can self host that feed an infinite stream of poison, although LLM crawlers are hungry creatures and would keep a % of your servers doing that
I'm curious about this. How are you doing it and what sorts of poison are you generating?
I understand that there are watermarking tools like Nightshade, but I hear they've already been undermined. I'd like to create a never ending rabbit hole of fake blog posts, but I'm not clear on how to ensure that the site's search indexing continues to work while poisoning any AI crawling.
The ones that say that Nightshade is not working is because they don't understand how it works. They "test" it by asking a LLM what image they see and it usually identifies it without any problems. The actual function is thay when the image is used to train data, it provokes errors in tagging the image. So a poisoned image of a car is correctly identified as a car by ChatGPT, but when is used to train the model, that car is used to train images of idk cakes.
For text there are a lot of interesting tarpits, like this one https://github.com/amenyxia/Sarracenia or the original one called Nepenthes
I'm not doing anything about "AI" because "AI" doesn't exist.
As far as the many, disparate technologies that are called "AI", I judge them each on their own merits like any technology. But it's important to be aware that capitalism ultimately develops and uses all technology for violent control. There's nothing special about "AI" in this respect.
Lastly I resist any technology that's actively oppressive whether people call it "AI" or not. It's difficult to talk about precise tactics because every technology is different with its own applications. For example, a company like Palantir should be stopped from infiltrating and exploiting our data on a mass scale. Similarly we should try to stop the automation of war and genocide. Etc. However these problems are nothing new and people have been trying to develop tactics for a long time.
I've been emailing/calling the politicians who are supposed to represent me and letting my voice be heard about my opinion on everything. From local permitting for datacenters right up to my senator letting them know how I'd like to see them vote on various tech bills and legislation I'd like to see sponsored.
I've been discussing it with the people in my life trying to advise them of the downsides to AI and pushing them to solving their problems through other means.
I've been pushing against it at work and in the open source projects I'm a part of to reduce use in them all where and how I can.
I've been helping to campaign some (mostly by being very vocal in my communities) for politicians I know don't instantly lose all rational thought when they hear the words "AI".
And, I've been doing as much as I can to make my family and friends more independent of the price spikes being caused by datacenters. Not full prepper but things like replace my old gas powered dryer with an electric one I can feed from solar panels and such (yes, I dry outside when I can too but I'm in an area where they'll get rewashed outside more than dried). We've all started picking up more hands on skills like sewing and things to help each other out (I'm the furniture repair man and gardener)
And on a fun note. I've been reviving my hobbies with friends that don't rely on tech so when they finally do force everything into rented cloud computing I'll shrug and roll to attack that dragon with my level 1 Paladin (yes, bad decisions were made by the party)
I'm refusing to use it. I'm pushing back on people trying to force me to use it (like my boss was trying for a while). And I'm pushing back against friends who pass slop off on me expecting praise for their ability to tell someone else kinda/sorta what to write/draw/whatever.
The best I can do against AI is prevent their crawlers from accessing my work. I've built iocaine for this purpose, and it's happily serving an infinite maze of poison to all the crawlers, and they ain't getting through. A number of other people and organizations also use it, and that gives me warm, fuzzy feelings.
I also wrote about surviving the crawlers, and helped a couple of friends have a li'l fun with them. That's actually quite fun.
I also helped our twins ween their teacher off of ChatGPT: she's otherwise a great teacher, but she's been using ChatGPT to give them homework. It had so many mistakes, typos! So they started to quietly correct the exercises with red pen, and handing the homework in, with the questions corrected too. Other kids saw that too, started doing the same. Then it spread to other classes. Two months later, none of the teachers use ChatGPT (or any other "AI") anymore, and the word of them being laughably crap spread to households that otherwise wouldn't be aware. New kid came to town, "you guys know about chatgpt?!" - he stopped talking about it by the end of the week.
Sometimes being openly, vocally very AI-hostile pays off. So I'm going to continue doing that.
Sounds cool as hell! Talk to me about Iocaine - is this for people who self-host or can I put this on my websites somehow?
It's for self-hosters, I'm afraid. The idea behind it is that your webserver / reverse proxy forwards all GET and HEAD requests to it, it does some heuristics and either returns a page filled with garbage your webserver/reverse proxy can serve, or a 421 (misdirected request) error code, at which point your webserver / reverse proxy can serve the real contents. So to run it, you need to be able to make it play nice with your webserver or reverse proxy, which pretty much means you'll have to self-host, yeah.
The garbage part is important, because that's how iocaine serves poisoned URLs (urls that have an identifiable substring), so if any of the crawlers come back, and manage to get through all other heuristics iocaine puts in front of them, if they land on a poisoned URL, they'll get caught anyway.
That sounds incredible (even if I had to google a lot of the terms haha). Would you do me a big favor and make a post about this in https://slrpnk.net/c/selfhosting if you haven't already? We would love to have you over on slrpnk.net, where we enjoy the blend of tech + the environment and hate AI. I would post myself but I'm just now investigating personal websites and switching to Linux so I don't think I'd be able to do your project justice.
I can do that, probably over the weekend.
Badmouth it behind its back. It's the only real thing that I got left to feel. Woops, I accidentally started singing.
But in all seriousness, what I am doing is to question people when they make it apparent that they blindly use LLMs. I ask questions such as
Sadly I'm pretty hopeless when it comes to individual action because of the insane amount of money involved. I had lost all hope earlier this year, but now there are more and more articles talking about how the math doesn't math for these companies' financials, so I'm now sonewhat hopeful that this would eventually put an end to all this.
I helped an AI project fail at work. TBF, it failed on its own, but I talked mad shit about it.
I also go out of my way to remove AI integrations from windows, browsers and other apps for my staff.
I am purposely checking in terrible code to github to make copilot worse.
Yeah, that's totally why. Yeah.
how horrible we talking?
Nothing. People want AI the same way they want TV and advertisements.
Also, software companies have been destroying software engineering for the past 20/30 years. I only focus on good stuff that I enjoy, small projects that still have a meaning to me and are useful. I'll write tools in Zig or Rust for fun, but that's all.
Once I retire, I'll go live in a cave or in the mountains and forget that it ever happened. Computers were magical machines to improve our life, but people don't deserve them obviously and they don't care about any of it.
THANK YOU for actually being someone in this community proactively wanting to mobilise others against AI instead of merely complaining about it! I really hope more anti AI communities will start to shift toward this type of discussion.
Anyways, something I do to fight gen AI is at my school, I keep a constant pressure against any AI generated content whenever possible. For example, when the school puts up AI generated posters seeking middle school camp counselors, I attach aggressive messages such as “cut the slop” and the “we need to kill ai artist” meme and immediately put more on if any are removed in order to keep up the pressure. Eventually the posters are removed entirely. (The posters are still readable and retain their original utility, but instead of making the school walls uglier, they constantly serve as a warning.)
I haven’t heard of anyone trying to sabotage any data centers so not enough.
If you want to sabotage one in ways that are really subtle and evil, apply for a menial job at one, and when they give you a tour of the site, drop rhizome fragments of Japanese knotwood wherever there's a corner with some soil that's a bit out of view.
That stuff fucks up concrete. It will grow into cracks and apply pressure at levels of a 25kg/cm², essentially exploiting any slight flaw in the concrete by making it worse. It grows into a large mass underground while above ground it only shows shoots at first. (People who know the plant will instantly recognize the shoots in question ... but how many people building industrial parks have that degree of botanical knowledge? And even if they did, how many techbrodude managers would take it seriously enough to tackle the plant with the alacrity and overwhelming force that's required to get it before it's too late?) That large mass will cause the ground to rise, breaking up pavement, making sidewalks and roads uneven, and, as mentioned above, causing expanding damage to foundations leading to destructive leaks and other such issues. It's an absolute monster of a plant and it takes forever to clear out. One missed rhizome fragment (as little as 0.7g) and the problem starts up all over again.
It typically takes YEARS to clear up a knotwood problem, at a price in millions.
And then you go back in and do it all over again. This time with clay packed with their rhizomes in the treads of your shoes.
Or you go back in and do it with any of the running bamboos of the Phyllostachys genus: Golden Bamboo, Yellow Grove Bamboo, Giant Timber Bamboo, etc. (The last of that list, incidentally, might even be scarier than knotwood for the sheer damage it can cause and the stunning speed in which it can cause it.)
Now if you do this, I will applaud vigorously from here on the other side of the planet. You're directly causing damage to the very infrastructure the new Nazis of the USA are building with these data centres.
Do not get tempted to use giant hogweed, however. That won't damage the data centre. It will cause permanent scarring and blindness to the working class folk sent in to clean it up. Focus your rage on the actual targets and their toys, not on their exploited hired hands.
Anyone know where to purchase Japanese knotwood rhizomes? Asking for a friend.
There's something very satisfying about the idea of destroying these "AI" plants with actual plants.