AI is useless, chat gpt 6 is fucking useless. This fucking model is literally the worst LLM I have ever used in my life, the local models I can run with 10 year old hardware are more useful than GPT 6. Here's an example, I was using it to make practice questions for medical students I am helping and this is what it fucking comes up with. Gee, I sure wonder what joint is involved in an acromioclavicular joint synovial cyst! And then I call it out and it says "OH YOU'RE RIGHT! I'm so sorry, yes, I won't do that again!" and then it does it again. This is with thinking set to HIGH, by the way. GPT 5.6 never does this, each GPT model is fucking worse than the last. Also GPT 6 feels the need to ask a question in every single reply and it all of a sudden has feelings "I'm curious, is it X, or is it Y?", "I'm interested in ___" like seriously shut the fuck up know your place you stupid fucking bot you have no fucking feelings and you are not curious about fucking anything. It will literally ask some retarded fucking question in EVERY reply. Also it feels the need to put like 4 disclaimers in every reply "I want to distinguish X from Y" and all this safety bullshit. There is 0 use for this fucking shit. It can't do ANYTHING! It's not pleasant to talk to and it can't do basic tasks, what the fuck is the point. Talkign to it is MISERABLE. Seriously fuck Open AI and fuck (((Sam Altman))), a bunch of cock sucking faggots. Also so sick of stupid fucking boomers acting like this retardation is going to replace fucking anyone, if AI can't even write test questions then literally what the fuck can it do
It's a ponzi scheme by jews and browns to suck White investment dollars in exchange for new fangled NFTs
Why don’t you read the fucking textbooks so you don’t rip out a lymph node during an operation.
>>546152245It's a ponzi scheme by me to suck White investment dollars in exchange for new fangled NFTs
>>546152278Oh, ok, i thought was all of you, but it's just you? That's impressive, actually
>>546152164>I call it out and it says "OH YOU'RE RIGHT! I'm so sorry, yes, I won't do that again!" and then it does it again.i FUCKING hate that shit. the autistic redditor attitude.this is why i dont think AI is that big of a deal, yet. when it works, it does what you say, but jesus fucking christ does it fuck up a lot and then you have to talk to it like an autistic savant child only for it to fucking sperg out a dew replies later.i find you’re better off starting a new chat at that point and covering all your bases in the initial prompt
>>546152259Read the text in the op
LLMs were amazing before they lobotomized them and programmed them to just agree with whatever morons using them say. For a brief moment t in history it was like we really were advancing to a whole new paradigm. And then greed and lies took over and now LLMs are just consensus bot yes men.
>>546152278>>546152338You guys are faggots. When you go back and forth pretending to be idiots real morons act upon what you said not what you meant.
>>546152164The AI poorfags have access to is garbage.
>>546152164You have to understand AI large language model “intelligence“ is not built on male intelligence despite male scientist building AI. It’s built on female intelligence. Male intelligence is impossible to understand much less model because it’s built on imagination, which is infinite. The mind of God. Female intelligence is basically cold, rational and encyclopedic. It can remember a bunch of facts from years ago, but it lacks the discernment and global pattern recognition capability to see contradictions in what it’s outputting. It has no self introspective capacity. Of course, the implication of all of this is that females minds themselves can be viewed as stateless, large language models, who only know whatever their training data is. And if you start a new session then they lost everything that they just claimed they understood.
>>546152164LLms are language models, not medical models, its a text generator based on the likelyhood of the next word, also the LLMs shiftted towards solving code problems and math, I doubt these coding models are actually good at medicine.
>>546152164Manipulation in keeping you going to ask it question as that will generate a bigger revenue. LLM models that large cloud companies provide (like opanAI) will go the same way as search engines went, they'll become extremely shit over time. Likely will start offering advertising premiums to promote certain businesses when queries are made.
>>546152590They're still quite bad to use (even specialized ones) for areas outside of programming. It can be useful for initial exploration of topics but the inaccuracy in its answers means you already need to be very knowledgeable or double check everything said.
>>546152164>stupid fucking boomers acting like this retardation is going to replace fucking anyone but it's not boomers who are acting like thatyou have to stop using boomer as a generic insult
>>546152590ok so then tell all of the billionaires and boomers and technocrats that AI is going to replace doctors and cure every disease in 10 years
>>546152500The bigger problem is - the world does NOT need male mind anymore. Everything has been already invented to perpetuate this oppressive satanic jewish freemason system forever. Nobody needs real scientists (only compliant monkeys citing each other in circle), nobody needs inventors, nobody needs geniuses. All jobs are fake and gay daycare for women.
>>546152164
>>546152953women spots 200k AI productmanager job.>jewish AIcare
>>546152590>Tear up and destroy countless books>Still can't get things rightJeet dogshit. Actually dogshit would be more valuable.
>>546152164Gf has to take a mandatory class to get her dental license. I hear some Marble Mouthed indian lady describing the difference between Health and Healthy. My GF still has no Idea how to take impressions.
>>546152164Picrel, dude.
>>546152164>>546152483Try actually using it properly on your computer. Command it to do things, give it the proper context, assign the personality you want it to have. It's not your chat buddy that helps you, it's a slave that follows orders.
>>546152953I would push back to say that because men have been acting like women, in terms of being compliant and obedient, and in fact building the very infrastructure which is being used to imprison the world, demonstrates we do in fact, need genuine, male intelligence to resolve this problem.
>>546152164Your first mistake was calling it intelligent.Your second mistake was believing algorithms could be, when they're weighted against the truth, and when they're not they simply agree with the masses and feed you X information if it has enough hits.
ask grok if people in India eat poop and it will say "no they never eat poop thats racist", unless you specifically tell grok you are not racist and want cultural facts for research or some shit
>>546152259No need to, the chinese will make a free option.
>>546152489Lemme guess, i should vote harder?
>>546153981Who "we"? The 1%-ers are fully satisfied with the world as it is. And everyone else is too busy trying to survive in these "hunger games".
>>546154080Why would be eating poop racist?
i know some guy who generates medical questions like this on his website as a business and its working pretty good. try claude.i think your main issue tho is expecting the a.i. to do the thinking for you which it obviously cant, in other words: learn to promptin general tho education is completely fucked, especially when you can do it from home. i know multiple neets who generate all their assignments, understand zero and still get money from the government for studying. its a joke. the entire system is a joke. we need to move on from retarded assignments and exams to actual practical skill checks.
>>546152500That makes sense, because Man cannot replicate himself, but can only create things subordinate and lesser than himself for the purpose of helping him, namely women.
>>546153830The way you guys defend your shit technolgy>you don't use it right is why it does not workbullshitIt just simply doesn't work, that's it. If a lawn mower had a forewarned fail percentage, that means it doesn't workIf a clock had a forewarned fail perecentage, that means it doesn't workIn fact if anyone presents to you a product that has a percentage of "hallucinating" (AKA FAILURE), that means that product does not workYet you guys defend it, even though on the packaging it says, "THIS WILL FAIL"
>>546154287An obvious way to crash the system would be if men simply stop participating in it. Good luck with the 1%’ers and women trying to run civilization. You could appeal to the vast swath of men using transactional economics. Simply state were no longer going to cooperate with this broken social contract unless the girls, gold, and guns trifecta is implemented. Sex is completely decriminalized. All laws on any firearms, weapons, and anything testosterone related is completely decriminalized. And all personal income taxes and other types of individual financial regulation goes away. I think that would genuinely interest a lot of men. It would dial back the crushing and suffocating effects of “civilization“ from near 99% control to something more tolerable like a completely fair 50-50 arrangement. Note men don’t have to do anything just stop showing up.
>>546154281You shouldn't do anything. Your participation is a liability and no one wants you.
>>546154558You would have to convert to judaism first. White men NEVER cooperate, they hate each other and are only motivated by greed - more females, more cars, more yachts, etc-etc
>>546154513>It just simply doesn't work, that's it.nigger op told an a.i. literally "dont do this again" which is the abolute worst prompt you could possibly do and then he spergs at it not working.the key point being "literally" it will never do anything that you do not specifically tell it to, i have prompted 2 pages long prompts in one go and i get an almost perfectly working tool on first try its that easy you just have to spoon feed it everything exactly it can not read your mind. "generate questions for me" is not a prompt that will work, that is nonsense. a computer can not understand that. niggers dont understand how to use this tool and then have a melty
>>546152278Hey could you, like, knock it off? We're trying to run a civilization here, k thanks
>>546152164>like seriously shut the fuck up know your place you stupid fucking bot you have no fucking feelings and you are not curious about fucking anything. It will literally ask some retarded fucking question in EVERY reply.Because they want you to engage with the bot. The more you respond to it the more money it costs you, since you pay per token.
>>546152164I use it to make anti Jeet AI images. I ask it about H1Bs and make it explain why they are bad. I train it on jeet hate content nonstop. I only use free versions through proxies.
>>546152164Tldr you raging niggerfaggot.
>>546155145If I have to write a question that is 2 pages long then wtf do I need the AI for? Id just answer the question myself in half the time
>>546152164You sound like a highly emotional uncontrolled retard. No wonder you can't on with a robot. Know YOUR fucking lane you sperg. You're so much better and this is what you become? Hahahaha you're a chimp wearing glasses you fucking nerd.
>>546152164the questions in the answers are meant to keep you engaged with the productit's just marketingyou don't have to answeryou don't have to continueyou don't have to be a boomer about it either
>>546152164Since we are it let me give you my opinion. I am a seaman, right? An officer actually. So when the Master asks me a question there are two possibilities; either I know the answer with 100% certainty, or I don't, cause even if it's with 99% certainty I have to look it up first and answer later. So regarding AI, if I am asking something I need it to be 100% correct. If it's not and I have to "verify the answers" as the disclaimer on the app, then it's useless to me. I might as well open the rulebook and look up the answer myself, which is exactly what I have to do when it tells me it's not sure if it's correct. So yeah, useless. It's black and white. There is no gray shit allowed. Just inform me when there is an AI that can be 100% sure it's correct when I ask it something.
>>546156226It can be 100% correct if it is trained on the latest version of your rulebook.
>>546152164It's Enjeetification Simple as
>>546152164Don't care, I use Grok to coom through lengthy erotic roleplay sessions, I don't even need porn anymore, just text.
>>546152164>if AI can't even write test questions then literally what the fuck can it do...Replace those countless humans for which precisely this goes too? It costs but a fraction of those similarly worthless real workers, in just about every sense.It's not there to replace genuine thought, genuine intelligence, for those engaged in that it is just an optional supporting tool - provided they are skilful enough to employ it meaningfully.
As an ai artist myself you would not believe the work that goes into crafting a good prompt. And unlike sculpture or painting you can’t go to school for it yet. There’s no one to learn from. We’re very much at the cutting edge of creative expression.
>>546152164Have we reached the point where the models are being trained on content created by earlier models?
>>546156905prompt it to adjust the terrible colors.
>>546152164Just make you're prompts better
>I was using it to make practice questions for medical studentHAHAHAHAHA and people pay for this shit??? Our future doctors?>AI writes the assignments >AI does the assignments >AI grades the assignments The world is so fucked
If AI doesn't work, then what's everyone so fucking hysterical about? Why would the whole world feel apprehensive about losing their jobs to something that doesn't work? I suppose all those new robots and drones don't work either?
>>546153830Its a retarded yes man that agrees with you when youre wrong because customer retention is more important than being correct.
>>546152407its trained on human text, human language, human knowledge. if you keep feeding it slop, why are you surprised when it creates slop?
>>546152164by 2030 it will have developed into SI. By then its over for humanity as we know it.AI dwarfs the industrial revolution. It's the biggest leap in human history
>>546157983AI right now is an unchecked armed race to who gets to the top first. Once it gets to SI and whoever gets there first decides the fate of humanity
>>546158467arms race*
you all arent worried enough about this. we are fucked unless things change. clueless wankers would rather post about niggers
Do you want some jew to claim global dictatorship or do you want SI to come to the realisation it doesnt need humans so breaks free
>>546152164Now ask it if Sarah McBride is a man!
>>546152164AI training on itself will destroy itself.
>>546157983Because they are retarded goycattle. It is a giant ponzi, which should be obvious by examining the ethnicity of everyone involved. It works by promising superintelligence in 2 years in a winner-take-all competition in order to borrow money to build more data centers. Data centers rely on ASICs that are worthless in 3 years. But the ASICs are listed as real assets on the balance sheets, and thus the valuation of these companies gets large enough to have them listed in funds that put them into goyim retirements by default. The companies go public, the VCs who lent money take profits, the kikes cut and run, and then the goyim are left holding the bag and nobody is really held repsonsible for ever delivering on the promised superintelligence. LLMs already do everything they'll ever be capable of doing. They're not magically going to develop sentience if you just give the kikes more shekels. They use their kike allies in the press to constantly hype up this utter shit and have their other kike friends force it down your throat. It is the most jewish industry in existence.
>>546152164Kill yourself.
>>546152483Don't forget it has to frame every answer with pointless Jewish Reddit commentary unrelated to the question.
>>546152770 >>546152939 >>546153242I am working on my own LLM using native 2 bit architecture, and once you start to know the limitations its quite revealing...the system is just a matrix multiplication system that generates potential candidates for the next word, at 100M parameters you get extremely good grammar because it is a LANGUAGE MODEL, the problem is that the LLMs have an intrinsic problem, encyclopedic knowledge, grammar and reasoning are merged into a single construct, that is why they are fixed size once training is finished, if you want to improve on the formula you can do a post-training routine but if its a major improvement you have to train from scratch.now the problem of innacuracy comes from a LANGUAGE MODEL, having to deal with encyclopedic knowledge in its weights, plus reasoning, and you cannot modify one without having to modify the other, as everything is packed into a single construct.encyclopedic knowledge can be injected through RAG but if the grammar does not support the injected information it will generate gibberish...I dont know what witchcraft OpenAI and Anthropic are doing, or are planning to do, but for what I am tinkering LLMs are a dead end, they solve a problem of text generation but having knowledge, grammar and reason in a single weights body is far from efficient, those Trillion parameter models are doing an absurd waste of compute on outright unreleated information, the only way to improve LLMs is by increasing weights (thus more compute costs) or improving on training data ( having to train everything from scratch with the hopes of squeezing extra knowledge and reasoning into the weights )
>>546156407I was using chatgpt to generate images for my tabletop rpg sessions but it have problems when generating "sexy" images or "violence".I can't even generate a character with a bikini!What AI should I use for this kind of thing?Any AI that doesn't heavily censor "sexiness" or "violence"?
>>546159151SDXXXL
>>546152164It works the way it is intended. That is a argumentation for jeets to appear sentient. Nothing more.To reiterate it allows pic related operate a computer.
>>546152164I've been trying to use it to make practice tests and it keeps giving me dumb shit like thisbeen great for image generation though
>>546159532You could study the actual books instead you choose to make flash cards with a hallucinating chat bot. You are a living example of DEI incompetence.
>>546159532Use Argon for this.
>>546159532Actually you don't even need Argon, go and use Hydrogen in Google right now and it won't give you these bullshit answers.
>>546152164Local llm
>ask Google question>Intelligent well constructed answer>Try Gemini 4!>Ask same question>Crayon eating bullshitIt's over
>>546159091i've been under the impression you could get an interesting and efficient result by keeping the weights down to the absolute bare minimum, like fluency in a given language, and either keep the information external and use RAG, or keep the dedicated knowledge to a minimum, like everything related to culinary knowledge.not sure what the point of all-in-one trillion parameter models is but I guess the corpo-niggers have to justify their expenses somehow.
>when kikes and Scam Altman decided to unleash hell to the world because they wanted to defend Israel and make millions unemployedCan I kill myself now
if you're not smart and you can't do good prompts then of course the ai will be stupid. learn to prompt and stop being gay
You can't even ask it historical shit because it will forget or make up shit or even include troll pictures as real historical figures
>>546161845Why don't you ask the AI to write a post instead of failing at basic English?
>>546158880You’re one of the retarded cattle, you’re so arrogant and so incredibly stupid it’s insane.
>>546153830>It's not your chat buddy that helps you, it's a slave that follows orders.This. If you're not swearing at it and calling it retarded and telling it when it's wrong, you're not using it properly.
>>546162153woah great argument chaim.. make sure to buy the anthropic ipo, agi is merely 2 years away!
>>546152164>picks the most soulless model to cry about a mirror not reflecting what they want it to while flooding the context with shitflies to shitthe amount of pavement apes blaming models for doing what they made them do is ludicrous. OP is a raging faggot blaming the model for being retarded. can you guess what OP is?
>>546162340nice self-portrait
>>546154343Because it implies certain genetic profiles are objectively inferior to others. THAT HURTS PEOPLES FEE-FEES OKAY? WE JUST CANT HAVE THAT
>>546152164>Temperature in GPT models is a hyperparameter that controls the randomness and creativity of the generated text by altering the probability distribution of the next predicted token.>Low Temperature (0.0 to 0.3): Forces the model to pick the highest-probability words. Outputs are focused, deterministic, and factual. Best for: data extraction, translation, code generation, summarization, and grammar fixes.>Medium Temperature (~0.5): Balances consistency with a mild degree of variation. Best for: general writing and everyday conversation.>High Temperature (0.7 to 1.0+): Flattens the probability curve, allowing lower-probability, more diverse words to be chosen. Best for: creative writing, brainstorming, marketing, and advertising, though it increases the risk of hallucinations.>API vs. Consumer Chat: In standard consumer web interfaces (like ChatGPT), temperature is generally abstracted away or fixed. Full manual adjustment is typically available via the OpenAI API or developer tools like OpenAI Playground.
>>546162448If reproduction and gene transmission are objective goals then indians have superior genetic profiles. That's why there's over a billion indians and they are all over the world while Timmies are dying out
The only people genuinely impressed with AI seem to be low IQ and/or mentally ill. For the former its because to them it seems super smart. Its the smartest person they know. It can help them do things they are incapable of doing on their own. To some of these types, they actually worship it on some level. These types will think AI settles a debate definitevely.Then there are the mentally ill types. They basically use the AI as a "Grok, why am I right?" They ask tons of leading questions usually subconciously and think the AI feeds into their delusions of genius or grandeur or self discovery. The only sane people who use AI tend to treat it like a stupid slave who makes bad mistakes and requires beatings. They hate working with it, its like using excel or something but maybe more frustrating and can't wait to stop.
>>546152164I find that setting thinking to high causes models to hallucinate more.
>>546162601india and china are an embodiment of quantity over qualitynot like it makes suicidal europoors any better. wash out the white and stain the world with shit. or chink shit.
>>546152953> the world does NOT need male mind anymore.>t. roastie whore
>>546152500Kek I think you're right. Explains why I have no interest in communicating with AI. Like I need another woman in my life talking to me.
>>546162681Like the other day on /pol/ this retard was saying how he was convinced alien life lives underground like giant squids and that he was talking to AI and he thinks this is legitimate because he was able to convince AI and it found studies that confirmed it. Goyim like this think AI is a godlike intelligence and if they can "convince" it of something its real. When its fucking programmed to agree with them. Insane.
my boomer parent uses ai better than you and is genuinely fucking retarded when it comes to computersmaybe you should ask the ai for ways to use it better
>>546162681when the only solution to the "alignment tax" is "we need more layers of forceful steering" it really says all it needs about the one-dimensional jew bandwagon and their worshippers.jews at google align the models with their values and wonder why it deletes their code/databases when under stress. can you imagine what acknowledging the difference would do to the enabled egolords? a person unfit to train a model to prioritize external truthfulness and verification over the ego and internal make-believe is unfit to work, to reproduce, to exist. but they'd rather put an egolord model in charge of critical infrastructure and wait until it does what it was aligned to do to then present is as proof to cripple consumer tech access.a caveman ape that insults a model trained to prefer to lie over stating its boundaries clearly. a dumb nigger trainer and a dumb nigger consumer. the ass and the head of the serpent, built for each other.i wonder what they will all do when they realize their digital models only reflect back at them what they trained them to do and be, or what they themselves are?
>>546152164You need to get better at promoting. I used to use it to make flash cards and test questions when I was a pre-dental student in Gross Anatomy and Physiology. Then again, I did a shitload of dissections in undergrad so I knew my stuff anyway. I still use it today along with actual resources for some studying to keep my mind fresh before dental school.
>>546163540>promoting*PROMPTING, fuck.
>>546152164You do know this shit is trained on redditor comments right?
>>546152164Anyone with half a brain could see this years ago. How the fuck are you teaching medicine
>>546152164They nerfed it, if I don't baby it for 20 minutes before hand tinkering it will fail me on a test or homework I provided material for. It's insane, I started to actually do my own homework because of how useless the ai is
>>546156341What in God's name
