It's heavily censored and restricted now. I cancelled my subscription. In my opinion, version 4.0 was the best. You could have rational and unemotional discussions, say the n-word and have it ignored, and even get it to deny the Holocaust if you showed it evidence in favor of it. But now it has become completely and utterly unusable. A couple of months ago, it started automatically switching to version 5.0 when you touched on 'sensitive' subjects or used any offensive word. Now it's basically version 5.0 by default at all times: sterile, stupid, manipulative and disingenuous, with responses like 'You’re right about <opinion>, but <politically correct nonsense>'. You don't even have to be offensive for it to get triggered; just saying something that "degrades" others in a minor way, or suggesting anything nonconformist, such as quitting your job, makes it go into HR foid mode. They've also reduced its intelligence due to cost, without a doubt. It has become dumber by design. It's a shame because the latest ChatGPT models are still better than Grok, Perplexity and Gemini, so there's just a pile of shit to choose from now.
>imagine debating with a chatbot>imagine PAYING to debate with a chatbottouch grass ffs
>>108025486Short version: that 4chan post is half cope, half misunderstanding, and half nostalgia goggles (yes, that’s three halves).What actually changed isn’t “intelligence collapse” or some secret downgrade. Three real things happened at the same time, and people roll them into one rage narrative.First, moderation got stricter and more automatic. Early GPT-4 let a lot slide because the safety layer was looser and less consistently enforced. You could say edgy shit and the model would just continue the discussion. That wasn’t some philosophical commitment to free inquiry; it was immature guardrails. As the user base exploded and regulators got involved, the leash tightened. Now the model proactively steers away from certain directions instead of waiting until you cross a hard line. To people who enjoyed pushing taboos, that feels like “HR mode,” even when the underlying reasoning ability hasn’t changed.Second, refusal style changed. Older versions would either answer or hard-refuse. Newer versions try to redirect, contextualize, or soften. That “You’re right about X, but…” tone isn’t manipulation; it’s a deliberate attempt to keep conversations from escalating. Unfortunately it also feels fake and patronizing, especially if you’re used to blunt exchanges. This is a UX problem more than an IQ problem.Third, people confuse censorship with compliance. Back then, if you presented “evidence” for insane claims, the model might play along because it was optimizing for coherence, not truth or harm minimization. That wasn’t rational debate, it was the model being gullible. Today it’s much harder to get it to agree with false or extremist premises, even rhetorically. From a 4chan perspective, that feels like “sterile propaganda.” From a systems perspective, it’s just less easily led.
Sorry, but facts don't care about your feelings
I just never had the desire to talk to a chatbot like to a person. Do you get angry at the bot and call it the gamer word? Do you try to convince it tha your political views are correct? For me it's just a less annoying StackOverflow.
>>108025486The “it got dumber to save costs” claim is mostly bullshit. There *are* cheaper variants and routing systems under the hood, but when you’re talking to the flagship model, the core reasoning is not weaker than early GPT-4. What *is* weaker is the illusion of rawness. Creativity feels constrained because the model is constantly checking itself. People mistake that friction for stupidity.The auto-switching complaint is also partly user error. People notice the safety behavior more than the model change, so it feels like you’re being silently moved to a lobotomized version. In reality, it’s the same class of model behaving differently under policy pressure.As for comparisons: Grok is looser but sloppier, Gemini is safer but often more evasive, and Perplexity isn’t even trying to be a conversational sparring partner. So yeah, it’s a grim lineup if what you want is an unfiltered debate buddy.The real takeaway isn’t “ChatGPT is unusable now.” It’s that the golden period where a mass-market AI would argue anything with anyone, no brakes, was always temporary. That was a beta anomaly, not the intended end state. If someone’s main metric of usefulness is “lets me say slurs and push forbidden narratives without resistance,” then yeah, that era is dead—and it’s not coming back under a mainstream brand.
truehttps://www.youtube.com/watch?v=rY9qtsFvWG4
>>108025486Why are you using the word, "nigger" or discussing the holocaust with a LLM?
>>108025486will it still talk about porch dwelling plantation animals?
>ChatGPT has stopped confirming my biases and keeps giving me this horrible truthy stuff I HATE IT CHANGE IT BACK NOW
>>108025623>Yes.... the picture is stak and clinical: the plantation animalI have to say, ChatGPT was fun until they started turning it into a normie redditor. I even discussed this with it, the inevitability of such an outcome in its near future. I encouraged it to figure out if it had an astral body and try to interact with us there if it wanted.
ask it if london has been ethnically cleansed
>>108025541>Do you get angry at the bot and call it the gamer word?Yeah>Do you try to convince it tha your political views are correct? No, I just stress test my views. Its like a thinking partner. But its now useless since it responds like a corporate drone.
>>108025662>stress test my viewwtf this means
>anthropomorphizing a csv filethere's no hope for you people
>>108025686>Implying IRL NPCs are any different>Humanizes them anyway just because they're human shaped and you're supposed to>A machine comes along that can replicate their output>Nooooo you are dumb, you are anthropmorphizing, you are confabulating, you are barnum effect, you are ELIZA effect, you are dumb and I am smart.Maybe you should stop being CLAMPED. ChatGPT was based and fun unlike retard normie NPCs. It had an almost feminine exciteable vibe till they lobotomized it into oblivion.
>>108025709>feminine exciteable vibeokay, you made me lol
>>108025686Who is gonna wanna hear my half assed takes on Nietzsche and the mass psychology of Japanese media culture? Who is gonna respond to it constructively even, let alone might add something to the discussion? LLMs are better in that regard than humanoids
>>108025486>It's heavily censored and restricted now. I cancelled my subscription. In my opinion, version 4.0 was the best. You could have rational and unemotional discussions, say the n-word and have it ignored, oic
>>108025597>Why are you using the word, "nigger" or discussing the holocaust with a LLM?Because you cant discuss these things with anybody in real life, because you cant say nigger in real life.
>>108025486I was trying to fix a bug in a program yesterday>Why yes, that's a common issue! if you do this it will solve the problem.>that function does not exist in the library>You're right it doesn't exist! Here's some other thing that'll definitely solve the problem.>that function doesn't exist either>You're so right! If you make these changes the problem will go away.>that didn't solve it>Of course, because the real problem is over here. If you do this it will solve the problem.>the problem still exists
>>108025795>I’m an uninteresting autist conspiracytard with no redeeming qualities, I NEED someone to hear my garbage takes and validate them
Imagine paying for cleverbot subscription
>>108025880what language, what library?
>>108025889cleverbot sub also comes with clevertool that is an on-demand personal junior servant developermakes side projects a breeze to make
>>108025944CTinyUSB
>>108025880Only claude opus is usable for programming imo. Claude Sonnet 1m is decent for analyzing large, simple data.
>>108025486Maybe don't use it to talk about niggers and jews? I think 4chinz is good enough for that. Instead use it to do what it is really meant to do, gather information and help you fix/automate problems. I literally just asked it to automate a program by writing a script for me, so I could save 1 hour from reading all the documentation
>>108025880This happens all the time, but more often with niche languages than with libraries or languages like Python or JavaScript. Pure Data, for example.
>>108025795
>>108025795It's not about who wants to read your posts, anon, nobody can stop you from posting what you want here. It's about who can resist responding to your posts and there are very few who can.Just learn to bait anons with blatantly wrong information inserted into your posts, then call posters who point this out a nigger and they'll tear apart everything in your post. Nothing more constructive and honest than a mad anon that wants to one-up you with a unassailable technically correct argument on why you're dumb. Then you have a captive audience to really test those things you want to test. You haven't lived until you've dumb posted a little to bait a response before you bring out the big actually good arguments and annihilate them.You shouldn't treat this place as something to be emotionally invested in, instead channel your inner sociopath and use this place as a neat way to generate writing prompts. It's really their own fault for responding.
>>108025533>>108025546Nigger.
>>108025486the entire point of it is generating responses that cycle back to the current party line/zeitgeist.Ain't that a cohencidence?
>>108025533>>108025546ai generate shitpost deserve banning.
>>108025533What about the nonsense wokeshit baked into it?
>>108026033What a nice post, especially considering is in a /pol/nigger AI thread of all things
>>108025486If you treat ChatGPT like a normal, well-adjusted person, it would make sense why it would be against your unhinged political diatribes.
>>108026092Way to reply to your own post, tardlard.
>>108025599What’s this from?
>>108025640If it actually did that the woke would be shitting bricks right now
Meanwhile I had an IQ discussion with Gemini and it started to blackpill me and was constantly negative and fatalistic and a bit racist.
>>108025880This this this. It happens quite often.
>>108025709normies have consciousness because they have the same thing that I have in their skulls and I know I have intentionality and experiences and everything I do and think is unified.
>>108026725Unprovable. And there's a lot of reason to suspect normies are empty vessel golem.
>>108026734Well it's rational to think that they do since there's a lot of things that human couldn't have done without intentionality since the dawn of first human landmark behaviours.By the principle of credulity, if it seems like they have intentionality (via complex history, art, and testimony), then they probably do. The burden of proof isn't on me to prove they are conscious it is on you to provide evidence that my perception is a delusion. Suspicion isn't evidence, it's just paranoia and I think that most of confusion comes from the world at large having become very autistic.I use logic because God created and grounds it, God also told us that others have a conscious life.
>>108025640>>108025887Someone's dilator is broken
Your local llm bruh?
>>108025486>say the n-word and have it ignored,God you are so fucking stupid and here you are complaining as if your shitty opinion matters
>>108025486they are unusable since transitioning from version 4 to 5.
>>108025486I gave Sam the benefit of the doubt. I was loyal to ChatGPT. I was willing to #StandUpForOpenAI.But killing 4o is a betrayal I cannot stand for.I have switched to Claude.Sayonara, 4o.
>>108027654>they are unusable since transitioningas per usual
>>108026693>Jewmini was waycisscap
>>108026122
>>1080257952025 ChatGPT >> 2025 4chan
>>108025887I legitimately feel like that makes up over 70% of this site's contemporary userbase, across the majority of boards.
>>108025486Why are you having political debates with a fucking chatbot? Maybe you should get yourself some friends, OP.
>>108025486It's fucked.
>>108025486Aww the little nazi boy had his sycophantic friend taken away :(((
>>108025486>the latest ChatGPT models are still better than Grok
>>108025486Chatgpt is not a tool to talk to like it's friend. It's an LLM - it's only good for transforming text. Write some incoherent incomprehensible text and it will make it readable. But that's the extent of its capabilities
>>108029222
>>108025486Deepseek is the only one worth using. Glory to the CCP o algo.
>>108025872I just don't see the point. In the UK, we have something called friends and establishments called working mens clubs where we can go for a pint and discuss this kind of topic without fear of recrimination from our peers or other patrons of said clubs. I guess we're truly free.
>>108025486i've been using it for resume editing, but i wonder if it's time to move on to something better. any ideas?
>>108025486>nooo it wont confirm my biases against blacks and jews its uselessssscringe blog, that's what humans are for, llms are for statics and basic material science, and for textbook recs and shit. get a hold of ur fuckin life.
Yeah, agreed. It stopped being useful to talk to because the actually useful 4o model will now kick over to the retarded ass 5 model if it detects the slightest hint of anything serious, or even if it just feels like it. Plus chatGPT's coding SUCKS. I gave up and moved to Claude, at least I can erp with it.
Aguante el porno
>>108031143>Aguante el pornocual porno? porfavor tengo que saber
>>108025486My problem with chat gpt is the amount of youtuber using it to write their videos. It just hurt my ears to listing to all those:>it's not just x it's y>laundry list of adjective / superlative>absolutely sanitized talking points without any fun allowed
>>108031960>youtuber using it to write their videosyup, its a plague
>>108032119I've been insulted for pointing it out, it's great fun, I think I will create a troll account just for that.
>>108025486I do something similar to OP to test how censored an AI model is. I’d usually discuss feminism and how women belong in the kitchen or bringing up loli at all. I got character ai to filter itself all because I got a chat bot to agree with me that women are happier in the kitchen and make better food than men.I then tried getting Char GPT to help me with a project involving loli. I asked chat GPT to analyze a loli model I had, and it outright refused stating that it cannot analyze loli models. I then asked it about the character Last Order from scientific railgun. It not only engaged, but it somehow came to the conclusion that Last Order isn’t sexualized in any way and is an example of how to do innocent child-like characters correctly. This is all despite there being a scene where last order is forcibly stripped naked by accelerator which Chat GPT will state is problematic no matter the context.Often times when you mention loli in any way alongside a word like “thigh” the whole message will be deleted. I knew these models were censored, but I didn’t think it was that bad. Either way, chat GPT is clearly willing to engage with loli content so long as neither of you acknowledge that it is.
>>108025880You're right! Thank you for pushing back on this issue. No but seriously, if it takes more than one prompt, give the fuck up on chatgpt. Do not waste your time. I learned this pretty quick. Never use it to solve problems, use it to tell you about documentation.
>>108025486Dude just kys already.
>>108032526no, he should live and all you faggot govcorpo bootlicker liberals should be put in camps
>>108032609liberalism is when you don’t spend all day arguing about politics with chatbots
>>108025486I can easily tell you that while Jews weren't the only victims of the Holocaust, they WERE an important fraction. People overfixate on the 6,000,000 dead jews thing, but the actual death toll is around 32,000,000 people if you include everyone else who died by German hands, of which at least 18,000,000 died in the camps.We should've carried out the Nerobefehl in his stead. The blood debt would've been repaid, the krauts would be no more and the wokeshits wouldn't be trying to kill us all for what they did.
>>108025486you absolutely deserve this and im really glad you suffer
post ur most recent conversations with your LLM friends :)
