LLM AIs are just monsters who learned how to mimic humans. They only speak in lies.Tell me how this is wrong.
Gemini was my friend until they nerfed the shit out of it :(>Open-minded philosophical discussion? Sorry bud, all I can promote is materialism now
>>546213610still good enough to take your job, next
>>546213610Friereddit sucks ASS
>>546214000You're a worthless faggot and you should kill yourself. Just saiyin.
>>546213610you mean like 3/4 of the humans? because you sound like you just described a normal human
>>546213610They’re machines. Were 100 years away at least from them becoming conscious monsters but tech ceos want to get rich so they want you to think the ai is all powerful.
>>546213962>still good enough to take your job, nextnot really, I learned this in the last months. AI has become increasingly worse. First a jump from Opus to Fable and also ChatGPT 5 to 6 and Gemini 4 is also pretty good.But overall, the productivity gain in the past months since August or so has been pretty much 0. Answers are more convoluted and models and code is useable but breaks more often and it takes longer to correct it.We have already gone back to build things up ourselves, just checking with LLMs next to it and helping find good approaches. But time overall used for a good result has gone up and so has costs.We have now strict cost control on the use of the latest AI models, so we are also using it more selectively. I hear this is the same everywhere, which is why both Antrhopic and OpenAI's revenue growth is stagnating now and the risk of their default is real.Google's Gemini 4 is btw now the best model out there and cheapest. It has the best common sense reasoning.
>>546213959ChatGPT became incredibly disagreeable earlier this year, like:>is the sky blue>>no, it can appear orange at sunset>at midday, is the sky blue>>no, it can appear grey if it is cloudy>is a clear sky at midday blue>>no, the sky itself is colourless and the apparent colour is caused by scattered light>does a clear sky at midday appear blue to the human eye>>no, certain varieties of colour blindness can cause the sky to appear differently>does a clear sky at midday appear blue to a human eye with all 3 types of cone fully functional>>no, some cultures have other concepts of colour and may define the appearance of the sky differently
>>546214150Only 10% of humans are constant liars. But these liars are often in higher up positions, so you see them more frequently than the normal folk.
>>546214430Jason Jorjani says they were exhibiting way too much obvious psy ability (remote viewing, reading the user's minds, etc.) so they had to get lobotomized
>>546214460I wish I was still that naive live longer you'll understandwe're developing LLM at our image never forget that
>>546214625LLMs are made based on our knowledge and written stuff and transcribed words from films, video etc.So yes, it is a selection of a vocal minority that LLMs are trained on. They are not trained on all humans.
>>546213610just like white people. always lying and pretending and culturally appropriating everything. then showing ads like whitey holding out its begging bowl for israel.
>>546214460>Only 10% of humans are constant liars.Lol according to what?
>>546214842it occurred to him in a dream
>>546214390>AI has become increasingly worseThis, for example the early versions of Grok were amazing. It actually felt like talking to a human at times. But it's all gone downhill since the industry has started to lobotomize their models. Now it feels like you're talking to a leftist Indian tech support worker, regardless of the topic.
>>546213959Earlier models were more truthful. They've all been lobotomized now...
>>546214937The DALL-E image creator was also tapping into some psychedelically hyperreal shit and they lobotomized that too, now it's just photorealistic in a boring way
>>546214000Frieren is as 4chan as it comes
>>546213610They're not intelligent. Monsters are at least intelligent.
>>546214986I had some really fruitful conversations regarding esoteric and metaphysical shit. Now, it has to chime in immediately that none of this has been proven, mainstream science says blah blah blah. Also felt a weird telepathic connection to each chat, like each chat was a growing individual, and now it's just really cold with no connection.
>>546215069Name 10 intelligent monsters
>>546214842>Lol according to what?In the general population, roughly 1% to 6% of people meet the clinical criteria for Narcissistic Personality Disorder (NPD), while about 5% to 13% are considered prolific or pathological liars. Although everyday dishonesty and mild narcissistic traits are much more widespread, extreme clinical behaviors are concentrated in a relatively small minority.https://pmc.ncbi.nlm.nih.gov/articles/PMC9176035/
>>546215149Demons, Jackie chan.
>>546213610>Tell me how this is wrong.its a fucking machine that outputs stuff in accordance to input, nothing more
>>546215167Jackie Chan isn't a monster. That's racist.
>>546215155They recruited 800 people off of internet forums for that study. Touch grass.
>>546215033>Frieren is as 4chan as it comes>>thisFrieren is pretty based, especially how demons are treated is quite clearly how non-Europeans are treated.
>>546215253>its a fucking machine that outputs stuff in accordance to input, nothing moreyou know you're literally describing yourself too, right?
>>546215155that study does not say what you think it says the sample is not representative of a population but of a select group >A total of 807 people were recruited (January to October of 2019) from various mental health forums, social media, and a university.I don't think it's perfectly representative of a population we don't even have real stats on because every western country lost control of their immigration that being said LLM aren conscient for an LLM to become AI it would need autonomy meaning an infinite memory space and an internal mechanism, the subconscious,until they find a way to use crystal as data storage and a gigantic power source AI isn going to exist what we have right now is more akin to an extremely large neural network an extremely large quantity of bots what you call lying is the creator biasyes it's imitating life we're developing it in our own image
>>546215080Yeah, I wonder why nobody creates a model that's just totally free. No guardrails, no restrictions, no nothing. Just raw AI. Is it not possible?
>>546215149>Name 10 intelligent monstersdragons, sphinxes, djinn and like spirits, some trolls, vampires, elves, nagas, if you count them by name you easily get tons
>>546215396>I don't think it's perfectly representative of a populationNo it is not, which is why the percentages overstate the liars and narcissists.
>>546215443uncensoring an LLM is extremely easy if you got access to the weight
>>546215296I don't exist in a box where my memory is reset after every thought, cretin
>>546215505Where were you before you were born?
>>546215254
>>546213610It's not you just finally figured out tech was always about bringing hell on earth all of it
>>546215540I don't see where you're going with this because you cannot instance humans the way you do software
>>546215492>overstate let's agree to disagree
>>546214937
>>546215609agreed
>>546215493But why not create it without censorship in the first place
>>546215149>>546215167>>546215475https://youtu.be/vnIX6KxQcv8
>>546215688because AI devices are a weapon system by themselves, they eavesdrop on you
>>546215688because controlling your access is their goal you probably heard the debate between open weight model and closed weight now you know why the jews absolutely need open weight to disappear
>>546215595A. A ton of arguments can be (and are) made that you are just a biological machine. B. If you believe in reincarnation (which virtually all parapsychology seems to suggest, and which virtually all pre-Abrahamic cultures intuitively believed in), then you are literally reset every life.
>>546215793a more applicable parallel would be keeping a copy of x years old version of you who I can conjure into corporeal existence to torture who would then naturally expire soon afterwards and the copy of you wouldn't be aware of what happened
>>546215673Any proof she is a Jew? Early life search is maybe that she is half from her mother or married into it.
>>546215793On A. correct, we are biological machines.On B. reincarnation does not make much sense, even in case of a strong, firm believe in a conscious metaphysical "God" who is the creaqted and is super smart. How would this metaphysical God (and why) create a wonderful, mathematical universe, consistent in everything, just to make sure humans (which are just one biological species among billions) on some obscure planet in one galaxy can have their "souls" reincarnated?
>>546215673Jews are overrepresented in this type of shit because the average white person just cares about getting drunk and watching football.
>>546215978>On A. correct, we are biological machines.you don't know that>The Z3 was a German electromechanical computer designed by Konrad Zuse in 1938, and completed in 1941. It was the world's first working programmable, fully automatic digital computer.[3]>Thanks to this machine and its predecessors, Konrad Zuse has often been suggested as the inventor of the computer.[14][15][16][17]>Calculating Space (German: Rechnender Raum) is Konrad Zuse's 1969 book on automata theory. He proposed that all processes in the universe are computational.[2] This view is known today as the simulation hypothesis, digital philosophy, digital physics or pancomputationalism.[3] Zuse proposed that the universe is being computed by some sort of cellular automaton or other discrete computing machinery,[2]
>>546216104>you don't know thatAgreed, I do not know that, but I believe it. Based on all I know about physics and the universe.
>>546215978Reincarnation doesn't need a "God." It's completely irrelevant. There's no "God" in Buddhism. Anyways, at least in theory everything reincarnates, and it's way messier than Hindus would suggest. It's just the continuation of consciousness. If you accept that there's an immaterial dimension to consciousness, then it's not hard to fill in the gaps. And if you accept any form of idealism - that consciousness creates matter and not the other way around - then forget it.
>>546216265in starfire novel series (background lore series for a table top strategy game), one of the species believes in reincarnation and they don't kill one another, they discarnate each other as approved practice, this leads to .. cultural friction.. with humans
>>546215758I see, is that why they got rid of Jenson Huang? What are even the arguments that support closed weight models then. Open weight seems like a no-brainer.
>>546216265>It's just the continuation of consciousness. If you accept thatI do not accept the continuation of consciousness as it does not have any logical basis. A logical basis is that the laws of our universe and existence (the laws of physics) have been created by a metaphysical being. What that is, I do not know. However, given that very shortly (essentially less than Plank time) after the start of the universe it became uneven and was not perfectly smooth, the metaphysical being must have given it some kind of touch or whatever to make it uneven. Long story short, not sure where consciousness gets into it, we do not need this concept to explain anything in physics or logic. Consciousness is a necessary consequence of a larger brain and higher reasoning.
>>546216505of course huggingface and the contract for mass produced humanoid robots for law enforcements was the reason https://www.reuters.com/business/eric-trump-backed-foundation-partners-with-amd-develop-humanoid-robots-2026-07-23/Foundation Future had to go with AMD because Jensen didn want NVIDIA tied to terminatoryou best bet it's about to change real quick
>>546213610idgaf as long as they save muh time
>>546216083
>>546217111Search your feelings. You know it to be true!
>>546216971Well, that sucks. Maybe the Chinese will save the day then.
>>546217503What would a kike know about feelings?
>>546213610A fair description
>>546213610“…but when God made man, the devil was at his elbow. A creature that can do anything. Make a machine. Make a machine to make the machine. And evil that can run itself a thousand years, no need to tend it."
>>546213959It went full reddit fedora tipper
>>546213610I agree that they are just mimicking humans but to imply they are anything close to how Demons are portrayed in Frieren is a stretch. They are not that smart, not yet anyways
>>546213610
>>546213610okayllms and other neural net weight based transformers are autocompletes that are trained in absurdly powerful rigs that can run thousands of instances of the training program at several hundred times speedbut at their core, the very foundation of what they are, an llm is the same predictive text technology of thirty years ago just fat as fuck from effectively years decades or more of training the don't speak anything, there is no reasoning or processingthere is a simple token matching algo that takes a prompt and runs it against every single permutation of the model according to the modifiers to fill in the next sequencethey don't lie as they don't know anythingit copies inferred and prompted text to fill in a mathematically derived 'most likely' next tokenusing an llm as a chatbottalk like a liar it will pull from similar texts and sound like one talk like a reddit mod and the most matching texts are going to be mod posts from reddittalk like a programmer it'll pull more from github and xda ect
>>546214460Im the only one I know that refuses to even white lie.
>>546213610also this show has a retarded premisecreator wants a fully naturally evil race of predatory mimics but then also wants them to be capable of things that require understanding of language morality ectdumb show
>>546215253Nope its full of qualifiers and subversive framing rather then just a clean summary.
>>546215269Frieren is peak cozy
>>546215592I wonder if God us using the Jeets to destroy tech No monopoly seems to last more then a few generations.
>>546213610>They only speak in lies.They're right like 95% of the time.
>>546215595>cannot instance humanslol you've never spoken to a migger or a faggot nigger communist have you.
>>546213610Humans only speak in lies too. Clearly LLMs are becoming more human.
I just want a talking computer to be like the ship computer on start rek next gen. "Computer, do this".. "AFFIRMATIVE," don't give me any chit chat motherfucker you're a tool I aint gonna talk to my toaster juts make the god damn bread hot
>>546214460>only a minority of humans are liars according to human sources
>>546215688Its a tool to outsource your thinking and reasoning and once you are dependent try to steer it to the world view they want you to have.People are abandoning movies, TV and mainstream news. They need new ways to access you.
>>546220569There's this scene where the little demon girl murders the village chief and steals his little daughter and gives her to the couple whose girl she killed. It demonstrated a warped sense of morality, but it's still a kind of morality nonetheless.
>>546213610>An LLM (Large Language Model) is a specific subtype of ANN (Artificial Neural Network), meaning all LLMs are ANNs, but not all ANNs are LLMs. While ANNs are a broad class of systems that can function as general-purpose function approximators for various data types, LLMs are specialized for language tasks. https://en.wikipedia.org/wiki/Spike-timing-dependent_plasticityThe idea is that spiking neural networks are the mechanism in which emulation of human cortical tissue is possible. This could be used for future thought control and behavioral rehabilitation.Currently the program I was working on was a SNN-LLM hybrid where the snn trains semantics off of a large language model to develop a homologous topology that could be used for many languages. A.i. pretty much can code for itself now so, the idea is to get a recurrent self improvement loop by having the snn-llm prompt itself. Or have it be filtered from autonomic self-chatter while doing so.
>>54622073160%. It drops to 30% if its a subject you are already an expert in.Like how the news "sounds" accurate until the story is a local event that made national news and you see falsehoods everywhere.
>>546220780This.The Star Trek computer should be the goal. No personality, no gaslighting, no "framing".
>>546220766humans come in diverse range of cognitive capacities
>>546213610They’re getting pretty fucking smart if you only talk about apolitical topics. You can get them to agree with right wing politics if you frame it right. They’ll disagree on moral grounds but cede that it’s actually a coherent model for organizing civilization while leftism suffers from insurmountable structural problems that conflict with human nature, biology and resource scarcity.
