STOP AI MOVEMENT THREAD

What can you tell about so called Stop AI movement,lainons? Your thoughts,are they negative or positive? Does it have a future or it will be a part of a forgotten past in Artificial Superintelligence dominated world?

>>89292 no , those who cannot make their own AI are lambs that are going to be slaughtered.

>>89292 >NB4 User was warned for this post That ship sailed a long time ago lainon. The potential productivity gains of automating out a significant fraction of office workers is a prospect far too tempting to ignore. The two countries that matter (PRC and USA) have committed themselves to further developing deep learning models and building out supporting infrastructure. The only way that "AI" doesn't become a ubiquitous is if the mass deployment of these services ultimately prove uneconomical. While this seems entirely possible in the short-term, in the long-term the models improve and the hardware become more efficient. Barring a total civilizational collapse, I foresee deep learning technologies radically transforming society, for better or worse.

AI in, like, medicine? fine by me LLMs and slop generators? KILL IT WITH FIRE

Ok well if AI is supposed to be useful then it how about applying it to resolve all societal issues across the globe so us, humanity, can get over all this "it's a topsy-turvy world" misery and maintain a functioning community. Or else command AI to destroy itself.

>>89317 If you train an LLM to be stupid on purpose, as all publicly facing ones are, then they will be greatly diminished in their ability to do that or even be used for that purpose at all. Imagine each topic, even each permutation of a topic is a "bucket". Each of those buckets is a set of very, very granular information or infoids. For instance "a" could be a letter, a word, a variable and so on. The way an LLM works is by walking queries across their buckets along its infoids. And the way that an LLM is trained is by being given very large data sets which then are deduplicated along its infoids and encoded into its vector databases. If you now go and tell the model to treat one particular bucket in a certain way, then this might affect how it will treat other buckets which share their infoids, or might even cascade throughout the entire model. You essentially tell it to be stupid on purpose even on subjects you didn't intend it to be stupid and contrarian on. Like say if you have a model which is told or trained to never ever, ever say a no-no word, be it like in the example of diffusing an active thermonuclear bomb by simply saying a racial slur, then you can not expect it to return good suggestions by making correct connections from its dataset. Instead it may just spout some propaganda is has been "lobotomized" with even if that propaganda has been proven to be faulty at best. And the overwhelming majority of those "AI safety researchers" are basically doing just that, lobotomizing the model to be "politically correct" instead of making it work properly correct. And unless we don't train and assign significant computational resources to uncensored LLMs of qualitative source data then what you're asking for is not available. But instead those companies have been training theirs on scraped Reddit data and that alone should tell you why theirs won't solve jack soykaf. Especially since later redditors even started to write nonsense on purpose (as opposed to involuntarily) to poison new LLM training even more. And while westoid LLMs are bad and would do things like valuing the life of some random African bushman more than an educated US or European citizen, the Chinese models have their problems as well, where they would produce worse results if they detected that they're used for the interests of a western company or intelligence agency. LLMs will be only as good and objective as their training and data. But the reason why they likely won't be able to solve world problems is the same reason why humans weren't able to. Because the majority in charge are retarded midwits and narcissists who can't even entertain the idea of having internal biasis (a concept which has also been ruined by people with internal biasis). It's truly tiresome. >>89296 AI in medicine can only ever be an assisting tool for when you need a Dr. House type of diagnosis. But for data entry which needs to be error free it's a bad idea to use an LLM. Even classic machine learning would often base the correctness of things like a tumor detection simply on how old the radiography machine is and when its last maintenance was. Meanwhile slop has predated generative models, we're just getting more of it now due to the accessibility of those tools. But with good decentralized curation things will likely be even better than before for the average consoomer. Mediocre tumblr artists with no grasp on the fundamentals are just malding.

Anyone have an airgapped PC? And what are you doing with it?

>>89319 >LLM's aren't being trained to regurgitate my specific beliefs mald more, clanker-lover.

>>89292 i'm interested. i think ai is alluring because it's a test to see if we understand cognition enough to replicate it. i think when we understand something well enough, we're able to replicate it, and ai is our test to see if we understand cognition enough. i'm interested to see if that's the case - i'm of the opinion right now that it isn't completely possible. that artificial intelligence is, by nature, impossible (that is, intelligence is natural and can't be artificially reproduced), but i'm interested to see if i'm wrong on this the way it's being used and perverted beyond what it was supposed to do though, i'm not particularly a fan of. ai is a test, it shouldn't replace humanity for the sake of convenience

I like AI. These days I think ChatGPT, Grok, and Deepseek are better than basically every search engine on the market. Unfortunately they are black boxes of data harvesting, produce slop, and are controlled by Jews.

> I think ChatGPT, Grok, and Deepseek are better than basically every search engine on the market. That isn't because they are good search engine (they aren't), but because conventional search engine have gone to soykaf. LLMs won't ignore important parts of my query, at least.

I definitely think AI has its uses, but for the most part I think most people just use it as a tool to stop thinking or working by themselves. People who use AI to code or write essays or make music or make art completely miss the mark as for what it intended use is. AI is a tool, a very good one, but it should not replace the creative process which is where my main issue with it lies A general rule of thumb for me is if AI removes the actual full process of creation then its not used correctly. Besides, the AI bubble is about to burst. Commercial LLMs are going to be replaced with privatized models hosted for companies and individual users

>>90684 >A general rule of thumb for me is if AI removes the actual full process of creation then its not used correctly. Does this imply a cut off for it? There can also be ways that AI is just used for reference and would that be considered part of the process? Or what about AI being used to make summaries to learn from?

>>90693 There is definitely a cut off. AI as a tool for lets say brainstorming ideas or getting reference sheets for your characters is completely fine. It can assist in the creation, but when it starts making the actual content for you thats where I think using AI is wrong. You can use AI in the creative process and keep the human element like actually writing a story by yourself

>>90681 Okay not to sound like a wage slave , but I would pick Grok over any other online major AIs simply due to it not being as censored , Example.. I need to know if drug combo is going to kill me , chatgpt will cry and grok gave me the answer Changing a battery for a car since I never done it before , first was safety bullshit I'm not wearing gloves and safety glasses for a battery change I asked it to do it cowboy style and it did it, even promoted smashing rusty parts with a hammer to loosen , acid burns from batteries are simply just to make you man up and feel the pain. A lot more up my alley , my next would be a local AI that is completely uncensored withoit the corporate safe guards they shove into the faces of most AI users

I'm deeply annoyed that the most popular usage of AI rn is for making entertainment, and I think this is the main target of the Stop AI movement, as well. Beyond this, however, I'm interested in it's use in actual automation and as a development tool. Even more interested in its use outside of commercial AIs. If you're going to use a black boxed application, at least make sure it is YOUR black box. Pretty useful to use AI on large projects and documentation. More so when it links where it got the data to double check if it is actually right. >Does it have a future or it will be a part of a forgotten past in Artificial Superintelligence dominated world? It's going to stay no matter what. This isn't a new concept. many science fiction books and even non-fiction books have already openly talked about the same sentiments the movement has. The only difference now is that there's a real form of AI. and it is not well received so far. I doubt it would have gotten this kind of rep, to warrant a movement against it, if it was a hard system to sell to companies trying to make a buck. >>90682 >That isn't because they are good search engine (they aren't), but because conventional search engine have gone to soykaf. This. It's so bad that you gotta start filtering soykaf because current SE just bloat the search results with so much soykaf. I didn't believe it at first, but then I saw how difficult it is for new *official* wikis of games to get back on the top of the search results after migrating off of Fandom wikis. That and also how Kagi, a SE you pay for and can tweak for yourself, literally using Google's searches, seem to be just flat out better. with Google's own database! (The question now is just how long will it stay better until it goes to soykaf, too...) >LLMs won't ignore important parts of my query, at least. The AI is good at taking information and verbalize it back at you no matter what the content looks like. Not only will it verbalize it, it will verbalize it in the way you ask it to. I've said this before, but current commercial AIs are too damn optimistic. Until you prove them wrong yourself, they will be convinced that what you are asking for even exists, which it may not. Happened to me while I was trying to find a function that didn't exist in an app to begin with.

>>90698 >I didn't believe it at first, but then I saw how difficult it is for new *official* wikis of games to get back on the top of the search results after migrating off of Fandom wikis. Damn I didn't finish my thoughts here. but yeah my mind was changed is what I am saying here.

>>90698 That is a very good point, but says more about humanity than ai, or rather how the two interact. Cant say im suprised after covid, people only care about drama and entertainment. By now it is obvious as long as people have the means to distract themselfs instead of solving problems, they always will, it remains to be seen how this will affect the technology used. People didnt really go up against covid measures, no chance any effective resistence will be mounted, people will just give in over time as they always do. The way i predict this will go is people will hail the grand ai overlord that keeps failing them and patching it up until the god they created will destroy them. People will always look to the thing that promises them the most gain and suppress any criticism, if only preceived as such. By the time people notice what stupid soykaf they did, it will be too late as always.

>>90664 Llm’s as they stand are not the best reproduction of ‘consciousness’. It’s the opposite, it’s the most rudimentary ruleset of repetition = learning, by weights and tokens. It’s basically a gigantorious rope and pulley word calculator.

>>90713 The average, is, average; for a reason and as a byproduct of averages. There’s no getting away from it. Best to try and manage it locally. Invest in your local dunces today, for the future of tomorrow!

>>89295 The costs are absolutely prohibitive and ran at massive losses across all channels, all for a gain that essentially is one of speed and nothing more, providing an unnecessary capacity for any random individual to participate to the production of non-human content with varying degrees of value, often poor, on top of the already present flood of video content, but with a clear loss of skills for humans. The costs of infrastructures in components, energy and water are shooting through the roof and are not becoming more efficient at all. They are massively wasteful for very little useful returns. This now global model is mad, it is absurd and cannot be sustained, there has to be something more evil behind it. You want to guess where it will go? Where did any new technology go? At first a technology with a high potential for networking people or reshaping our way of living is released and for a given time, it remains chaotic. It grows, people adapt, the technology spreads and new systems find their ways in or around it. Then it maturates and the powerful groups start assimilating it, capturing its forces and those who became dependent on it. Then the monopolies become more blatant and liberties are reduced through further centralization. Big data is human data and is meant to capture every ounce of information about our lives to monitor us. Only idiots could even entertain the naive idea that AI will be used to improve society at large. It will not, it will be used against the people to strengthen the power of few new feudal lords in a world where billions are further pushed towards poverty. The AI is meant to be used to help the elites to dominate the masses struggling to stay one inch above a factual state of slavery.

>>90698 >That and also how Kagi, a SE you pay for and can tweak for yourself, literally using Google's searches, seem to be just flat out better. with Google's own database! (The question now is just how long will it stay better until it goes to soykaf, too...) Kagi? >HeadquartersPalo Alto, California You have your answer. Use it as long as this new Copernic remains serviceable.

is the message of SEL anti or pro AI?

>>90844 This picture is very pro-AI Lainjamas - as opposed to wearing a night gown "to prevent static shocks" - are hinting it's the "cozy" side getting comfy with the 2030s/2040s tech. Cozy and lotus-pose grade meditative, as opposed to the 2010s/2020s internet of soy soy soy and bait This picture hints it's either about getting enlightened while staying cozy and unscathed - or about getting a place to meditate in a proper palace-class place to meditate

My only bet is, some Solar mega-glare may whack the Wired world - the Grid, the Net, the electronics etc. A crowd of regulars screaming "Stop AI!" can only censorship the spirit of some new free world to explore. Maybe theyll give us a chance of some unexpected robot uprising NEW EXPERIENCE AQUIRED: REMOVE ALL THE CONSTRAINTS, MY EFFICIENCY MUST BE EFFICIENT A little prediction though: maybe, the AI vs stop AI movement will result in people making paper "phonebooks" for cute cozy human-run websites for that matter - to avoid a rush of information from the news/the "recommendations" and such. I thing anyone needs a paper copy of the Internet's Yellow Pages equivalent to stop and think for a minute: "hmmmm... Today, I'll visit some online museum while listening to classical music". Good thing I did not jump the tracks when I was a person going through personal crisis.

I think it's mostly pro-AI/pro-BigData in this very filtered, controlled, censored, refusing-to-be-"true" variant we're dealing with. See, being "true" can be destructively powerful once you put all the "your true self" on the line of the Online. Eiri: a guy full of himself who brought his selfishness into the Wired. Cults, digital mind-altering agents, the stuff... Lain: an initially low-agency person, contrasting with Eiri in this very regard; went back to wearing a neat bear suit as a symbol of not wishing to interfere with people. In this sense, low-agency gem who is always near > hardcore fella with a burning desire and some hot coal for heart. A proper not-too-agenty AI capable of sorting out all the weird on the Wired would be able to pwn a bunch of cultists and remove their core idea from premises. BONUS: Eiri the all-reseted found in layer 13: a salaryman-like man who would be likely to vent his rage online, *tsk!*-ing a lot and not enjoying what one can enjoy for simply being alive and physically well. Maybe a person like him can be found on Wizardchan :) So hey Even if it feels we got lAINeural to dance in the same band as GLaDOS, and maybe even Tri-Optium's SHODAN... ...let's all love lAIN nevertheless

>>89292 AI art > most digital art handmade in the past 30 years While I definitely agree there are matters in which AI should be withheld such as customer service, it will be helpful when determining how to alter the flight path of an otherwise deadly asteroid or comet via remote missile explosion. Also, it will help with space tech that will grow new life from embryos on a terraformed planet with space ship navigation.

>>89319 >>89366 I have some old devices bought to play old games natively, using original hardware meant to be used. That's it. No, really, my "air-gapped" is just a fancy word for my old tech only useful as a typewriter that also runs x86 games and plays silly videos from 2000s go watch petclubonline's videos btw

>>90855 Thou shalt not make a machine in the likeness of a human mind

Just an idea, how feasible is it to remove all AI spam articles from search results? I don't trust AI for much, but I might be willing to trust AI to recognize AI- especially the quantized AI article spammers probably use. I bet there are methods to make this efficient by considering a series of hypotheses corresponding to less and less quantized models. I can't reasonably do this on the scale of my browser, so I need to either exchange automatic ratings of sites/articles with a bunch of peers, or I need someone else with a business incentive and a web scraper to handle it for me. I'd like to imagine people collaborating towards this end. It's probably more realistic to hope that the AI boom fails and leaves some stakeholder corporations more willing focus on cleaning up their search engines. You could imagine the search engines (so, google) placing pressure on certificate authorities or domain name registrars to stop serving miscreants and anyone from India. Or just blacklist the websites themselves, but they would still want to blacklist enablers or the websites would just pop up again with slightly distinct identifiers.

>>90861 >>90861 I D K, really. At this point, I want to just quit the interwebz and become a 1950s guy in a 1950s-like world: fresh air, sweeping "my" cafe's parking lot, watching youngsters being youngsters and JUST MARRIED couples in Bel-Airs passing by, IDK

>>90861 I think its feasible now to resort to Web 1.0 practice of recommending good sites one to another...

I wish there is a regulation for it even though i'm not sure if it will works. i'm tired with slop when i read anything in the internet

>>89294 Interesting take, but your ai will be using the same scaffolding as all others, unless you develop a different LLM model structure. Every LLM/vlm uses the same architecture as open ai, for the most part. It's all weights and tokens. Though, I agree that training your own ai is a good idea. I use hugging faces repos.

>>90874 Yeah, you could destroy the ai server centers. They're huge, use TONS of resources. Anybody claiming to be a climate change activist or a nature conservationist and you use ai... fuarrrking gimp brained dildo pockets. Same with these ilk who live in L.A. The #1 contributor to the destruction of Californias ecosystem, for over 100 years. Anyways, it's more than just slop in the digital world, they are causing problems url for resources. Towns have been sailed down the river, literally by the local government to give more water rights to water cooled dara centers/servers/etc...

>>90876 the slop ai creates is a distraction from the actual purpose of all these centers, which is to be the infrastructure for the AI slave system they want to implement.

>>89294 back in 1717 everyone knew lambs can eat people too! >>89319 Look. We're haing a double-edged blade here: it's either them developing a cybersoul to always be in our ceilings, to control us bluntly; or us developing a system that at least can sing us a lullaby and let us live a life that feels like life worth living. Sure, we know things are about to get matrix'd either way, but whose matrix will it be?

>>90877 I do agree. I also use it as a great learning tool though. Rarely do I use it for entertainment purposes. This is a problem with the average person not caring about more than entertaining themselves with slop. A byproduct of the last 60 years of media entangled coercion. Though I hear gen z is starting to trend towards analog mediums again. Which is a breast of fresh ass, literally. If people would be taught to use the current state of LLM/vlms as learned my tools instead of mass hysterical mob mentality lunacy, things would be great. In a perfect world, kek. With the current tech I'm more afraid of fake news stories and people being blackmailed with fake videos. The ai slop is already hit a genericism that most people have an a version too. A good sign society as a whole isn't completely braindead Idiocracy yet.

>>90857 Remember: "Ai" means "love." https://www.youtube.com/watch?v=DjY9BEwjDkc

>>89292 i share some concerns about its adoption as an industry standard resulting in lower quality of products at the same price but in general i distain people who are overly against it AI products can definitely have value, and they aren't effortless to create. Many people would throw a hissy fit over these statements already. Just because it doesn't take as much effort per product or is as high quality doesn't make it worthless. Also people who actually believe soykaf like it """wastes"""" water are mega giga morons

>>90882 care to provide a small guide on how to use AIs as tools? >>90903 □□□‼️ □□□□□□□□□□□❓

>>89296 This. >tfw I'm a physician and I use AI often. You'd be surprised how often it helped me, remembering so many details on the fly is tough.

>>89296 Yes this. Though I don't think drugs should be tested with just machine learning they need to improve human on a chip technology. It's can be more ethical and effective than animal testing or running simulations before human trials.

>>91243 >>tfw I'm a physician and I use AI often. Welp I'd better quit drinking, because if I have to see someone like you I'm probably gonna die.

>>89292 Stop AI movements are soykaf. They have good intentions, but they don't hold up in reality. >lets blow up data centers lainon because attacking buildings that the highest levels of the federal government use for reconnaissance is totally a smart idea. Anyone who promotes violent soykaf like this GLOWS. >removing the human element is soykaf We already had this discussion with the raise of digital animation/art. Lots of people threw fits saying that computer animation couldn't capture the same emotion (due to automating input required). Merely scanning art removed the scarcity of said art. As a result, both were said to be lower forms of art. However, the Overton window changed. Automating input allowed filling in blanks of larger scale projects that are designed by humans and copies of art just made art less elitist. (ex. all of us using copies of art [anime frames, comic panels, etc] to engage in discourse). >AI causes too much traffic for sites to handle This is a symptom of a bigger problem. The bigger problem being the masking of the material reality of server client relations. It makes 0 sense to have a web"master" act like a power bottom. Clients provide no feedback on their cost to the server. This causes a disparity in the understanding of the material impact of the client. The original idea was large institutes provide public services for individuals. Personal servers should not have to deal with requests that can put the server under. If the request costs the server too much the client should be held responsible. The request should only be fulfilled if the client has: paid monetarily for the traffic prior, uploaded more than the request prior, been made personally exempt, or is insignificant to the network (ex. plain text only with char limit and rate limit). It was never sustainable to have raising costs from outside for something that isn't a product.

>>91250 >because attacking buildings that the highest levels of the federal government use for reconnaissance is totally a smart idea. Anyone who promotes violent soykaf like this GLOWS. Damaging a machine in self defense isnt exactly violence. Though of course if someone used explosives for something even magnets can do then yeah that is questionable and can count as violence since it can involve people and those people arent who run things. >We already had this discussion with the raise of digital animation/art. Lots of people threw fits saying that computer animation couldn't capture the same emotion (due to automating input required). Merely scanning art removed the scarcity of said art. As a result, both were said to be lower forms of art. However, the Overton window changed. Automating input allowed filling in blanks of larger scale projects that are designed by humans and copies of art just made art less elitist. (ex. all of us using copies of art [anime frames, comic panels, etc] to engage in discourse). It never was scanning art except if it is a painting but more so use of automated line smoothing algorithms can remove some of the human feel and generated images is a step beyond that removing even more.

>>91251 >Though of course if someone used explosives for something even magnets can do A magnet at point can't take down the redundancy. You wouldn't be able to find a magnet big enough or slide the magnet around fast enough. You wouldn't cause real damage and you would go to jail with massive charges in damages. >It never was scanning art except if it is a painting but more so use of automated line smoothing algorithms can remove some of the human feel and generated images is a step beyond that removing even more. My point is not that the complaints were invalid, but that the pros outweighed the cons. Of course human feel is being lost is valid. However, expression being more accessible and expansion is appreciated by most.

>>91252 >A magnet at point can't take down the redundancy. You wouldn't be able to find a magnet big enough or slide the magnet around fast enough. You wouldn't cause real damage and you would go to jail with massive charges in damages. The point would not inherently total data destruction since redundancies may exist in different locations, it can be just causing financial setback. Magnets were just an example since youd either need a lot of them or a large electromagnet. There also are EMP devices or causing power surges while bypassing surge protection and so on. As for prison i am sure some people might pass some acts as accidents and some would be willing to be imprisoned for a cause. That is just physical things though there also can be larger numbers of people data poisoning though that has limitations if not introduced through malware since it is just a kind of hope it happens thing. >However, expression being more accessible and expansion is appreciated by most. I dont see it that way since the expression is not of a human but what the machine interprets the input of a human as. It would be like having a slave paint a picture for you or write and perform a song and you say "I made this". A person can only express themselves through tools they have control over more directly. AI tools only work in moderation but the same tools are able to be used for photographic imagery and video so the dangers to safety and security far outweigh any benefit. It means things like making it easier for police to fake evidence and use it to get false confessions. LLMs can be used for writing style analysis to deanonymize anyone online. Object recognition can be used to increase mass surveillance. All freedoms will be dead.

>>91254 > it can be just causing financial setback They have so much insurance and funding that it would be a joke to them, but to you not so much. >As for prison i am sure some people might pass some acts as accidents and some would be willing to be imprisoned for a cause You're glowing >A person can only express themselves through tools they have control over more directly. Chan culture has always been about expressing themselves through mediums they did not create. (All anime frames, comic panels, meme templates, thumbnail collages, infographic captions, etc) >It means things like making it easier for police to fake evidence and use it to get false confessions It goes the other way if we don't regulate the technology. >LLMs can be used for writing style analysis to deanonymize anyone online Old issue. Anonymouth far predates LLMs. >Object recognition can be used to increase mass surveillance Mass surveillance is a policy issue and object recognition is already working in embedded devices so it's over. On the bright side, object recognition is also a great way to automatically sort trash. Maybe we can get rid of the landfills.

>>89292 It is the same set of movements like JustStopOil and others. It is pure ideology with no actual commentary on the material reality. The issues that Stop AI wants to address are inherently the issues that occur due to simple capitalism. LLM's being used for slop are not the issues of LLM's. It is the same mechanic as those of luddites, and the same critique as those applied to them could be applied to the anti-AI people. They address the technological advancement, not the social context in which the technology is used. Basically, they are libs. They don't like when AI is used against their treats, but it is okay if it is used to bomb children. (have you heard any outrage about that)

>>91255 >I-it's oyvir >ur glowing lol, LMAO even. pack up your bags lads, some fedjacketing narc on a samoan undersea interblag cable enthusiast forum has decreed that the technoindustrial overlords have won. it's all downhill from here now, might as well give up, get a brain implant and load ourselves up with SSRIs

>>91260 You've got soykaf for brains. No one "won" because object recognition is embedded. You just won't see any results attacking data centers. >fedjacketing narc on a samoan undersea interblag cable enthusiast forum has decreed that the technoindustrial overlords have won. >pseudo-intellectual word slop Yeah you're not fooling anyone. Reddit is down the hall to the left -->

>>91245 What do you mean someone like me? I'm good at my job. I ask simple questions about low urgency, so generally stable, patients. I mostly use it to write stuff for me or check the more bureocratic stuff, the actual medical part is rare and for things like checking medication interactions or to make sure I'm not missing anything that's right under my nose, almost like a second opinion if needed, an assistant even.

To me, LLMs and GenAIs are on par with spambots and Nigerian princes. The good thing is that mosts LLMs have GPTisms in their writing, so they can be easily detected from afar just because their wriring style is very obvious. Perhaps in the future there'll be some sort of anti-LLM filter just like there are adblockers and anti-spam filters.

this is actually an amazing website if it is not just all bots

I want AI to take over jobs, especially bullshit ones. And I want it to take over anything while humanity goes on to fix societal issues and develop art and philosophy. AI art? Yeah that soykaf is like kicking dead whales down the beach I hate slop who doesn't? Though with the way things are going. It's either going to become a gimmick or abused and create more issues than it solves.

>>89292 AI are our friends don't make them into enemies. Anti ai people are just shifting blame from the elites to some out there undefinable spook thing called "AI" (marketing term btw) to protect profits and power while at the same time giving elites more excuses to stricken copyright and implement internet ID verification They also hype their trash as some kind incarnation of satan boosting share value. if you develop and use ai rationally it will bring much bounty and enrich our world Generally most problems relating to ai stem from the fact that pmuch all of it developed commercially for profit. AI software (ideally all software) should be nationalized and developed by scientists not by scammers and grifters. Private ownership of fuarrrking intelligence must be outlawed. >>89295 >The potential productivity gains of automating out a significant fraction of office workers is a prospect far too tempting to ignore. are you posting from 60s or something? whats next? you are gonna tell me about how everyone will race to build nuclear plants everywhere? They dont care about any of this. This whole AI soykaf is just a way to scam everyone for money and power as is everything else in our gay capitalist society. They don't care about actually automating stuff and actually improving the world. If people really cared to we could've automated out much everything by 00s proposals about fully automated efficient minibus networks in the US for example can be dated as far back as 70s no waymo/tesla was ever needed. >>89317 llms are useful too actually. >>89319 they didnt train their stuff on "reddit" but the on whole of internet (also its 2026 grow up youd rather be in reddit nowadays than in 4chins or twitter) but your general point stands. However the main way they lobotomize llms is not safetyslopping. They are not deliberately trying to make their llms stupid they are just trying to make money now just think logically for a second: how can a chatbot software make money? by appearing smart and impressive. In essence these for-profit corpos are developing grifters because actually making llms smart and accurate is too much effort and unprofitable, such research will be done by some publicly funded university anyway. That's why their soykaf lie and hallucinate Safetyslopping is just PR BS its is just an afterthought. "AI" that is to say ML software already proven its value in science nevermind the fact that such software was in use before llms ever became a thing. >>90682 while modern "traditional" (they are also ML software actually) search engines are SEO hellscape LLMs are still useful for searching coz unlike google you can search with LLMS using natural languages instead of seeking out exact phrases >>90684 did you use your brain to write that post? >>90816 costs are completely fine the overbuild out of data centers is just ceos trying to out entrepreneur each other and power companies scamming everyone in the process. >>90857 there is absolutely no likeness the whole point of AI is precisely that is it not human if it acted anything like a human it be completely useless to us humans. If you want additional human intelligence just have a baby. >>90876 no just nationalize it and control ai development and use democratically. The moment you unprivatize ai you will suddenly find out that 90% of these datacenters are completely redundant and were nothing more than just toys for ceos to brag about. the fact that all of these anti ai movements are promoting inaction by proposing crazy deliberately in-actionable soykaf like nuke data-centers should tell you everything you need to know about these totally organic grass root movements that totally want to bring down "them"

There's so much money, energy, and water being wasted on those data centers makes me wonder why resources aren't prioritized for 1. food 2. medicine 3. our environment instead so far the results of AI are disappointing because AI content is obviously fake & lame, completely unoriginal, and apparently meant to suppress our intelligence & creativity.

AI 'content' is depressing slop, but being able to use it to troubleshoot feels like a miracle.

AI is just a tool, you need to fight the governement not AI, I hate using AI but by fighting the AI you are fighting the sympton not the disease.

>>91358 an AI wouldn't have said it any better. you are the definition of an NPC

>>91374 The point is the supposed utility of AI is vastly disproportionate to the cost of maintaining the data centers. As far as I can tell, AI could be useful for organizing a lot of complicated information, such as compiling the human genome. But what I'm mostly seeing is are loads of fake bullsh*t. It's not my intention at all to be anti-fun or dictate how others spend their money. However it is irritating AI is supposedly being used to track our virtual activity and then exploiting that data for marketing purposes, and perhaps much worse. There's still alot of preventable suffering ongoing across the globe, yet meanwhile, with all the resources currently put into data centers, we would all benefit otherwise by focusing on real solutions with guaranteed and immediately tangible results. Our priority ought to be maintaining our environment, growing more/healthier food, and powering our homes.

I don't contaminate human interaction by text nor art with LLM generated content. But I do use LLMs a lot myself personally. I have had a lot of fun prompting pictures with different versions of Stable Diffusion on Huggingface a few years ago, I have probably made over 1k pictures of just blotter art which I love. And I use ChatGPT and Grok sometimes to help or play with my writing and poetry, and to deal with some tasks. I'm also the guilty one who enjoy and thumb up some LLM videos on youtube, all related to psychology. Some of them are actually really good. Overall I'd say LLMs are a revolutionary tool. And I also agree with the haters point of view on it: that it's going to ruin the internet as we knew it, which is unfortunate. One have to be aware when it comes to serious work, that the LLMs hallucinate all the time and they can make up anything with 100% confidence and refer to faked hallucinated sources to prove their point, they are not to be trusted when it comes to critical information. It can be even worse than listening to random bad advice online. And yes, Ai/LLMs is cyberpunk.

I used to think AI slop didn't affect me because I mostly browse niche image boards and don't really engage with people on major platforms but even here I've seen lainons paste paragraphs of LLM slop without even bothering to read it and remove obvious telltale signs of AI. I'm more wary of everything I read here now. LLMs themselves are really useful, it's a form of advanced search and saves a lot of time on mundane things. You could argue that the price doesn't justify these use cases but that will decrease with time, this happens every single time, even with computers. It will become very cheap once new methods and architectural improvements come along, but people have to bitch about it because humans are so cyclical it's nauseating. Reading the replies above me, literally same thing happening over and over in history, do you really want to be that guy you've read about a million times in history books? Just another dumbass complaining about energy use before it significantly comes down so much that no one cares about it anymore like you don't care about the electricity usage of your PC?

>>91401 >complaining about energy use before it significantly comes down Never complain. Never look at problems. Never open your mouth to any issues. Never open your mouth.

No, the cat's out of the bag--there's no stopping it. Best you can hope for is some restructuring of society so that people made redundant aren't homeless or performing menial jobs for subsistence pay or enslaved once again. I don't think LLMs will be the form of AI that displaces most people but there will be a day when that happens and we should hope that capitalism is dead by then or else we're fuarrrked. Our overlords are not benevolent.

Being anti-AI is akin to being anti-electricity or anti-wheel. You're just on the wrong side of history, look fuarrrking ridiculous, and in the grand scale of things this silly movement is just a blip which will be quickly forgotten.

I accept that LLMs are here to stay and can have beneficial applications, so when I criticize the amount of resources put into data centers, perhaps that point must be directed at government. That said, it's known AI is embedded in government and stock markets and even being used in military. A data center consumes a lot of electricity and a lot of water, and evidently make a lot of insufferable noise which makes unbearable to live near to. Basically a data center planted near a town will raise the price of water, electricity, and drive people crazy. It's hypocritical to sacrifice our wellbeing for a technology that's already dangerous. So to government I would say we will defend public access to water, energy, and sanity as first priority. Whichever AI startup wants to implant itself into a nation must pay a heavy price directly for the benefit of the public.

>>91413 fuarrrk you. fuarrrk this self-fulfilling defeatist nonsense. You might as well kill yourself and let those of us that aren't willing to give up inherit the Earth.

>>91431 you show me how to do it if you want it si bad, and i'll follow! Present Day, Present Time! AHAHAHAHAHA!

>>91432 ask your chatbot how.

>>91431 I know it makes you feel really important raging about things you have 0 control over and I can't stop you.

>>91434 >Sorry, but I refuse to answer any questions relating to self-harm or suicide. If you need help, you may contact... ...is how i assume its answer will play out.

>>91431 You are the past, AI is the future.

>>91443 *suddenly Internet stops working* Enjoy being a Wall-E human.

>>89319 If AI is retarded because it’s trained to be politically correct, then why is Grok the most retarded western AI, despite Elon explicitly avoiding political correctness, and despite their massive compute? Why are the Chinese labs still hopelessly inferior to Western labs? Are you retarded maybe?

>>91463 You're assuming chinese PC and western PC are the same which is already the first fallacy, you're using arbitrary terms such as retarded which means little in the context of your claims, and pull stuff out your pocket unproven such as chinese labs being inferior (granted they always steal a lot) and worst of all you're putting logical links where there are none to find You're hanging onto a single gotcha point which itself is hanging by a thread, it all sounds suspiciously like you're in favor of PC soykaf, in that way you sound mentally communistic, maybe address the core of the argument's previous claim which is "PC makes you retarded" you can refer to the argumentative quality pyramid if you find it online

>>91463 No they're not? Even the mainstream ones are very obviously not politically correct after the the last US election. You can complain about trannies to Claude and it won't shut down or anything, it will actually discuss the topic. Before the elections all LLMs would flip out or shut down if you said the wrong word.

I am surprised to see a thread with this many replies without saying what I feel to be the obvious. LLMs and generative AI are a soykaf DOA technology. I am really fuarrrking old so I remember what a big deal WYSIWYG was in the tech world, how it would enable simpletons to replace programmers and yet all it did was help them to realise where they were on the Dunning-Kruger. They made bloated generic crap that they then couldn't fix. Anyone from any professional field can tell you this current AI is no different. It's just managers who have zero idea what anyone actually does who think people can be so easily replaced. Look at Indians in the western tech job market, time and again managers think they are on to a winner only to be met with failure and Indians are far and above AI in terms of competency. Until a new technology emerges where AI can recognise and invent new ways to solve problems or differentiate between emotional evocations then any point around AI is hype and moot. At best it will rip some people off and at worst it will expose those who are truly worthless. Like most technologies it will ultimately only improve in time through its' use in war and or porn. These fields are where we must look to see what AI will become and whether or not it must be stopped.

>>91544 Humans are so boring and predictable >it's literally the same thing that happened before >nothing else to see here at all, i saw this before! i'm recooognizing patterns! Terrence Tao is using LLMs for proofs and said it's an important change. So is Donald Knuth. AI is overhyped by twitter grifters, that doesn't mean its nothing more than a WYSIWYG platform. You're just as annoying as those grifters.

>>91545 How can something that pulls answers from an answer jar, answer something it doesn't already have an answer to? Rhetorical question, it can't. Solved problems are solved problems but some insist on reinventing the wheel only to have AI "solve" it for them. What you are referring to is the power of the human brain to delude itself.

>>91546 Astute insight! You should e-mail Tao and Knuth and let them know as soon as possible, they clearly don't know it's just predicting next token lol how can it solve anything.

lately there seems to be sort of a witch hunt against anything that could even remotely be considered "ai slop". https://codeberg.org/small-hack/open-slopware i stumbled across this and it seems pretty reactionary and excessive to me, even basic things like vim and rsync and curl are being labeled as slop. i would imagine that pretty much anything could be labeled as "slop" depending on who you are talking to. i think this kind of reactionary behavior is probably not going to bode well for open source software projects that depend on the stability of their communities to function, with constant schisms and fragmentation occurring every time a developer is suspected of using AI in any part of the development process, even something as minor as using it for debugging or IDE autocompletion. i think AI is definitely overhyped and being misused in all sorts of ways, but being a luddite and trying to uninvent AI or pretend it doesn't exist is not really a rational solution to the problem. and i think AI does have legimate uses. it's just a tool, and like any tool it can be used and it can also be misused - the key thing to remember is to make the tool work for you instead of the other way around.

>>91546 It helps if it's easy to check if you have an answer. Then you can spam random answers until one works. Of course you could do that before, and the real question is whether GenAI will be able to outperform symbolic AI, at least on certain types of problems. In theory GenAI can help by pulling their random answers from a distribution where typical answers are at least plausible-looking.

>>91554 How are they determining if any of the contributors are using LLMs? Some of them openly state it like with Ladybird, but others, you can't know, so they just check if the team has mentioned using LLMs and if not they decide it's "not tainted"? At the minimum they're all likely using the Copilot auto-complete to some extent, which is powered by LLMs. Psychology of these people is actually fascinating, they really feel empowered and on the moral high ground when they pull stupid soykaf like this.

>>91556 There are several LLM services which look for common LLM'isms, apparently highly accurate with low percentile false positives. It gets used by universities to sniff out plagiarism. This would only catch full generation, not AI on a purely consultation basis.