The paper uses a mathematical compartmental model (a type of nonlinear dynamical system) to simulate how a population interacts with LLMs over time. The population is divided into three groups:U: Non-users.C: Competent users (those who use LLMs but retain independent thought).D: Dependent users (those who have offloaded their ability to think to LLMs).By setting specific fixed variables (represented by Greek letters), the authors simulate how people shift between these categories.1-0 humanshttps://arxiv.org/abs/2609.03344
>>110044690>when you're a retard everything looks like covidtrust the science chud
>>110044690>AI fans have a transmissible mind diseaseIf we keep going at this pace, all of my schizo hunches are going to become credible. How long before they figure out most people are actually just biological LLMs?
>>110044714Go to bed, Richard
/g/ is mostly situated in the D group it seems nowadays. Pretty sad to see a bunch of talented young men degrade themselves like this.
>>110044941nah, its mostly tourists from the shartythey were fucked even before ais were a thing
>>110044690Here is the conclusion in simple terms:1. The big idea:LLMs spread like a virus — not biologically, but culturally. Through friends, schools, workplaces, and platforms copying each other.2. 3 kinds of users:Uncoupled: little/no LLM use, uses books, web, own thinkingCoupled but autonomous: uses LLM plus still reads, writes, checks, thinksDependent: lets the LLM do the thinking for them3. The danger is a tipping point:Adoption looks gradual, then suddenly flips.Once enough people offload thinking, independent thinking gets harder to sustain, which makes more offloading attractive = runaway loop.Result: abrupt drop in what people can do without the AI.4. Lock-in: prevention is easier than cure:After the flip, just going back to old conditions is NOT enough to recover.You have to push much harder to get autonomy back. That's hysteresis / technological lock-in.5. It's not "AI bad":Key difference is scaffolding vs substitution.Scaffolding = AI as tutor that helps you learn.Substitution = AI does it for you, skill erodes.Example: "think first, then ChatGPT" preserves creativity, unrestricted use hurts later unaided performance.6. Cognitive immunization:Not avoiding AI, but staying resistant to dependency:keep unaided tasks, verification, note-taking, critical discussiondeliberate disengagement periods, keep non-AI skillsdesign AI in schools/work to support, not complete, the taskif severe dependence: self-regulation and even therapy may be needed7. Caveat:This is a simple math model, not a prediction. Real life has networks, continuous degrees of autonomy, and co-evolving human-AI agency.I've been saying this. We need to stop the slopping machines right now desu
>>110044941>Talented young menDon't kid yourself the furries migrated long ago
>>110044964FPBP
>>110044964Did you just ask an AI to summarize the paper for you!?>>110044690>competent usersSounds good. AI has some good points that can help you at some tasks or make you harder, better, faster, stronger, but so does cocaine. How can you make sure you’re staying at a safe dose?
>>110045022Yes.
>>110045050Weird flex, but OK.
LLM(s) are the Thing from John Carpenter's movie.
>>110044986>FPBPSI dependent brainlet detected
Did this account for everyone in D thinking they're a C, only about 0.1% of Cs being actual Cs? How can you trust your own judgement in this case?
The first country to ban AI/LLMs and ban social media will be the next global superpower. Screencsp this.
>>110046244Afghanistan?
Useless academics trying to stay relevant not realizing that power and control now rests on a few ai labs who don't care about the whims of the average underclass.
Does this also make rich people retarded?
>>110046391Oh definitely
>>110046412Seems to me it isn't very different from their life in general if they have humans and assistants handling most of their shit
>>110046391Yes. I have to deal with them and they're the dumbest motherfuckers in the world. Literally Somali their.
>>110046464Guess that explains a lot about the general state of human societies over time.
>>110046391Define rich.You mean "rich" like Elon or do you mean old money rich like the actual decision makers of our society?
>>110046583Old money mostly. Seems like they've lost the plot.
>>110046451just watch how Silicon Valley operates, they pump money in tech startups that may one day become profitable and not just one, a whole ass portfolio. Most of which doesn't generate them shit and is a loss, they're all gambling that one of them works and becomes the next facebook or something.
>>110046223>Did this account for everyone in D thinking they're a C, only about 0.1% of Cs being actual Cs? How can you trust your own judgement in this case?Yeah, I see the problem. Suppose instead of asking /g/ to spoonfeed you, you asked ChatGPT to read the study for you. Does that elevate you from a U to a C or do you plummet to a D? Come to think of it, what's the difference between a U and a D?
This puts what I've thought all along into words. An incredibly large number of words.
>>110046223>>110046701But to answer your question: if you've gone as far as your own thinking can take you before reaching for the LLM, you're a C. But then again, what if you're so stupid that your thinking was never really getting you anywhere in the first place? Are you a C or a D? Something about this study doesn't check out.
>>110044964I asked my chatGPT to operate under a more Socratic method and guide me and ask questions so I can figure out the answer for myself rather than just give me the answer. It works pretty well.
>>110046989>i can totally figure out the answer for myself>it's just that i need the bot to ask me leading questions and guide me instead of asking my own questionsCategory D.
>>110044690Are we there, cyberbros?
>>110044690I'm clueless, but when I ask a question on chatgpt, I don't immediately believe the answer given, I always look for many websites for comparison and check the information to see if there are any contradictions or not.With the way I use AI, I should be considered C
>>110047056If you're not a D why are you asking the shitbot in the first place instead of thinking for yourself or at least reading the relevant material for yourself?
>>110046748The thing is that growth comes at the point of resistance.
>>110044941Unfortunately AI technology innately provides the biggest capability boost to low IQ third world ESLs.
>>110047101You have to know english pretty well to even prompt properly
>>110047097Then push against your internal resistance and try to figure out the obvious response to that instead of me issuing it.
>>110044713>covidCovid did cause dramatic and permanent change:Every business suddenly realized they can get by with doubling their profit margin by providing the worst product / service and pricing out a third of their consumer base.Wide spread retardation is the equivalent hidden cost of AI, not the mainstream unemployment and AI doomstay propaganda the slop merchants peddle themselves.
remember how using logical search operators and actually using your brain to think how to capture the specific information you were looking for and filter out all the other chaff was a thing
>>110047114>know English properlya native could not prompt it without a table of supported linguistic and semantic operatorsyou overvalue your birth statusa chink doesn't have to know what a tadpole is to prompt ithttps://github.com/aj-geddes/useful-ai-prompts/blob/main/prompts/supply-chain/supply-chain-excellence-director.md>muuuuuuh native status>--ACK!
>>110044690it's basically propaganda to suggest that C exists. that's like saying that you can curate an instagram reels feed that helps you with your career. im sure it's theoretically possible, but to think a human could manage it is a fantasy.
>>110047081time efficiency, even though I dig up a lot of information from places like wikipedia, reddit, youtube, it takes time to find answers to my questions, it would be efficient if AI summarized it for me even though many of the answers are hallucinations
>>110047190And even if C did exist how can you be sure you're not sliding into D over time? C isn't a permanent state.Taking the U route is by far the safest option.
>>110047192>muh efficiencyTake your assigned seat, D.
>>110044964Kinda defeating the point of automation if you're still doing all of the stuff yourself. I think a better mental model is exercise versus manual labor: we used to live in a world where hard labor was required from every man of working age. Physical labor is important to a person's health and well being but once industrialization comes along it doesn't make sense to keep plowing fields by hand just because your body needs exercise. So we started doing recreational exercise for well-being and because being strong is still generally useful even if a machine can do 99% of the physical labor we once did. So with competent AI the correct response is not "let the machine teach us how to do the thing we already do but better, still do it by hand" it is "outsource the work to the automation machine, shift skill development and maintenance from something that happens as you work to something you do for its own sake"
>>110047168>remember when actual search engines existed?Ooops. Can't show that in a capitalist manga.
>>110047304>we just need to shift skill development from doing the things that build the skill to uhhh... hurrr... ummm...>it's just like muh heckin' recreational exerciseI hate these "people" so much it's unreal. The global elite are doing nothing wrong. Cattle exists to be milked and then slaughtered.
>U>D>CNew /g/ meme just dropped.
>>110047304You'll outsource thinking and improve your other brain?
>>110047311sure can't.
>>110047168Remember all the garbage websites you have to sift through to look for relevant information?
>>110047398I remember sifting through interesting websites to look for relevant information, zoomie.
>>110047334Disagreeing is fine but you should explain why i'm wrong instead of being a vaguely superior faggot.>>110047381I can work way faster by automating various parts of my job, I'd like to keep working faster while not losing the ability to do those parts of my job. So I'm going to spend some of the time I'm saving explicitly practicing those skills.
>>110044964@grok summarize this100 characters max
>>110047475>you should explain why i'm wrong instead of being a vaguely superior faggot.Ok. Give me something more concrete than "something that happens as you do for its own sake", such that it's meaningfully distinct from "something that happens as you work". Then I'll explain why you're wrong without being vague about it. Deal?
>>110047512"something that happens as you do something for its own sake" *
"I'm a C."t. D
>>110047512>I can work way faster by automating various parts of my job, I'd like to keep working faster while not losing the ability to do those parts of my job. So I'm going to spend some of the time I'm saving explicitly practicing those skills.Let's just start here
>>110047571>I can work way faster by automating various parts of my jobYou're getting automated out of a job. You can forget about jobs for the purpose of this discussion.>explicitly practicing those skills.Ok. How are you going to practice your skills? What are you gonna do? 15 fizzbuzz pushups in the morning? Solve leetcode?
>>110046451congratulations, you have discovered why there need not bee an actual conspiracy for the world to bee in the state it is in
>>110047762Conspiracy denialism is a mental illness.
>>110047771proper orthography on furchins is a sign of much worse menthol hillness
>>110047672In the future? Absolutely. My job is uniquely well suited to RL environment construction, it's just a matter of the labs throwing enough compute at it before it's totally automated. I keep one or two personal projects to work on entirely by hand, stuff that interests me. I just do the parts of my job that I enjoy.
>>110044964>consuming crack without becoming addictedgood luck. our brains are hardwired to take the path of least resistance.
>>110047838>I keep one or two personal projects to work on entirely by handTo do what, now? WORK on?
>>110044964I asked my Claude pretty much the same question and got the same answers as listed herehmmm...
I already feel AI induced brain atrophy so I'm going to quit it cold turkey.
>>110044964same with television, once people waited for the men in suits sitting at the fancy tables to give them their conclusions they stopped making their ownbut that's a good thing, eveyone being on the same page means homogeny and peace, no infighting until that UNHOLY internet threw a spanner in the works, but they're working on fixing that as we speak, it will soon be in lockstep with legacy media
>>110047840realistically, it's the only path forward without the entire world collapsing on itself because the next generation is literally braindeadbut perhaps that is the planthere's a reason everyone who is rich enough is building bunkers...
>>110047853I spent about three months offline last year and it felt like I had a whole new brain.
>>110047858Are you seriously "Wild West Web" posting on /g/? The internet is their tool and always was. Even the Fidonet guys knew it back then.
>>110046583all of them are retards completely enslaved to their biological impulses who also think they are 200+ IQ>t. knower
They should have a fourth category : non-users who have had their ability to think severely impaired due to anti-AI hysteria.
>>110047914Massive cope.
>>110047920I asked chat to analyze your response and he said there's a high likelihood that you belong in this category.
>>110047952I asked Raptor Jesus and he said that your brain atrophy would be visible on a CT scan.
>>110047883did hiro ever pay that OFCOM fine
>>1100479854chan isn't a problem because its good for a regime to have it's discontents on the outskirts of society because normal people avoid them.
>>110047846Work is done for the purpose of economic value, this is done for the purpose of skill development. Does lifting weights in the gym count as work?
>>110048005convenient.
This thread proves every C is a coping D.
>>110048014Yes. It's the same with the Left being dildo obsessed weirdos. No one cares about their critiques of power if it comes wrapped up in the degenerate shit that goes on at Pride.
This is all just smart people who are upset that being smart doesn't make them special anymore.
>>110048040Ultimate LLM-hate trvke.
>>110048040>AI is good because it'll make everyone as stupid as meWeird flex, but okay.
>>110044690not my problem, because I never had cognitive abilities. I never had my own opinions or independent thoughts in the first place.
>>110044964AI completely fucked schools already. It's a cheating device. Kids won't learn anything and I saw this happening already a few years ago. They just let a LLM think for them so they will have free time.
>>110048206oh no, kids these days sidestep the retarded busy work
