>>24493439 (OP)
I will not trust AI until I am able to input anything I want and control all the censorship cockblocks on my own. Until that time comes, I will not trust it.

ChatGPT is fucking retarded. I ask it simple questions and it gets them blatantly wrong. Then I noticed everytime I correct it, it just agrees. So I started “correcting” it with blatantly false information. It still agreed. These AIs are pretty much useless imo. They’re basically just a chat bot toy.

>>24494443
>These AIs are pretty much useless imo. They’re basically just a chat bot toy
That's exactly what they are, a chat bot toy. A lot of their 'training' comes from fanfiction slop, blogs and reddit posts. Then the actual LLM models were developed through customer support chat bots and coomer chat bots like replika.

>>24493439 (OP)
No. GPT is the worst for /lit/ purposes.
It goes Claude>Deepsneed>>>>>Gemini>>>>>>>>>>Local Shitter>>>>>>>>>>>>>>>>>>>>>>>>>>GPT

GPT is tuned for stemcels.

@grok rephrase this in hindi

>>24493439 (OP)
You used Google for everything, this is no different.

>>24494451
Claude is best for lit? Really?

>>24494597
By far.
Deepsneed has the best one offs by far, truly masterful to the point its scary in some cases, but it is unstable and falls off or gets stuck in loops of repetitive or bland writing after 4 or 5 messages. Makes it unusable given the amount of work you have to put in to original prompts to get a single gem out.

Claude Opus is known to have the best writing throughout by knowers, and it consistently delivers high quality.

Gemini is meh but stable for extremely long context windows, like novel lengths.

GPT is lifeless slop - always.

>>24494619
Interesting. I've been playing with Gemini 2.5 and have definitely been impressed, but since I don't have a Pro account, I haven't pushed it. I've actually never tried Claude.

>>24494633
Long term the best thing is to switch between models/providers/jailbreaks as they are all significantly inbred in terms of training on their own data, and once you feed too much of it’s own responses back in the form of chat history to the context, slopisms of each model get reinforced until it gets stylistically locked into bland stupidity.

Write 50msgs to Gemini, then switch to Claude or deepsneed, 50 more, and switch again, and so on -if you’re trying to write a book.

>>24495877
Are they really all that close to each other? I mean, generally? If true, how does Google not have far and away the best position in all of AI, for the average person who can’t afford to pay for 3-4 services?

>>24496516
I'm by no means an expert. If you really want to learn about it go consult the AI coomer ephors in /aicg/ or /g/. They know all of this better than anyone.

By consult i mean lurk. Asking them anything or for any kind of help will be met with scathing ridicule.

>>24496549
I heard. All I’m really interested in is if I should be paying for ChatGPT Pro, Gemini Ultra, or Claude Max. I’m almost exclusively interested in what they can do as a creative enhancement and for everyday tasks, and almost nothing to do with anything more than basic coding. Google’s pricing is high but if they throw in all that other stuff, it’s hard to deny their offer. I’ve been using ChatGPT from the beginning but I’m just not seeing much from their offer that I can’t get elsewhere, but I will stay if they’re the best.

>>24494443
>ChatGPT
This. It can sorta help, more so for some editing or seeing obvious errors like a missing period or something like that. But beyond basic edit checking, it's kinda shit. It can't write for shit unless it's corporate business speak, and quite frankly I wouldn't want it to write for me. It does a shit job at creative writing but it can kinda give some interesting suggestions and it comes to similar plot conclusions and ideas that I have when trying to plot out the next conflict in a story.

>>24498149
It really just reflects the person using it.

>>24493439 (OP)
If you even use chatgpt a little bit as someone who isnt sub 90iq you would quickly realise just how fucking dumb it is

>>24493439 (OP)

>>24498171
Again, it reflects the user. It's like complaining a sports car is slow because you don't know how to shift gears.

While I admire the thought and the “commitment to craft” or whatever, some of you are in for a real rude awakening if you think that people are just going to not only discover your work but that you’re somehow going to convince people that yours is “pure” and untouched by AI or something. Or that they’ll follow you and continue to support your commitment to your asceticism rather than just go prompt their own story. Truly, I admire your naïveté. But honestly in for a rude awakening if you think this shit is in any way going away.

>>24498181
Retard. When it gives the wrong answer or contradicts itself two messages later it is simply dumb.

Not to mention how hard coded it is to be politcaly correct.

>>24499842
It’s only ‘retarded’ when it doesn’t agree with you. Classic cope.

>>24493439 (OP)
I believe in it more than the university of 4chan where people post meme images to share information

>>24499935
Im talking about it being objectivly wrong about history often.

>>24500072
As if there's only one version of history you fucking retard.

AI has inspired me to—wait for it—keybind the em dash to my keyboard.

@grok is india lion yes or no

>>24499935
>>24500087
AI doesn't really just reflect the user. This is a recent thread with a moron: https://warosu.org/lit/thread/24492095
Whose chatgpt says the exact same but the user is a complete moron. Contradicts himself multiple times, is objectively wrong several times, and gets gang raped by 5 anons, all while memeing chatgpt is 10,000 IQ

the only reflection is that anyone who would use chatgpt is a moron and will shortly become one. https://arxiv.org/pdf/2506.08872

>>24500172
Still seething about that thread? LMAO. That anon destroyed you fags, that's why you were begging him for a redo.

>>24493439 (OP)
It's only good at a few things, one of them being to pull the wool over midwits' eyes.

>>24500087
>one version of history
https://copilot.microsoft.com/shares/yLmcHNQyybmGWs3DUApmG
How about completely fake quotations. I primed it for a fake one, but when talking to gpt users on the internet I frequently word my posts to prime their models to fuck up and call them out after they fuck up. It takes them a long time to realize what is happening, and will often ask their model something like
>Is this quote real?
And it will say
>Yes, it is found at page X
They won't verify it, because why would ai just get it wrong?
When pressed and they try looking for the quote, they'll ask
>I couldn't find the quote, where is it?
The ai will say
>Due to differences in translation,...
And the ai user will continue on the same trajectory. You can notice malicious wording in people who argue with chatgpt intending to fuck someone's model accuracy up all the time now too.

>>24500195
>copilot
Maybe try using something else you dumb fag.

>>24500201
I used that because it allowed quick sharing a chat without an account. You can easily do it with any other model.

>>24500204
So it's a skill issue, got it.

>>24500205
Are you 12?

>>24500207
Only 12 year olds use fag AI's like copilot.

>>24500210
ChatGPT has turned you into an unironic moron.

>>24500219
What? Are you okay?

>>24500223
That's ChatGPT giving false information. Here's what it said when I asked it to dig deeper to find the quote. This thread is about ChatGPT being used to self-learn, I'm answering the OP, it's not good. Why are you getting angry and defensive?

>>24500230
It's a tool just like Google, it's not a God.

>>24500236
It's an atrocious tool for learning, and seems to result in lower brain function over the course of its use, and has an impact on understanding regardless of how it is used overall. This isn't strictly known yet, but here's the study: https://arxiv.org/pdf/2506.08872
The previously mentioned OP was using ChatGPT like a god rather than a tool, so he got turned out and fucked up. Ideally someone would take a step back and reassess how they use arguments, but given his tantrum and failure to address the arguments I'd guess he'll dig himself in out of pride.

It's sad, but that's the general trend for education and why this board is getting so full of GPT jeets and the illiterate.

>>24500256
You lost, get over it. The fact you're still seething tells us everything.

>>24500087
Ah yes thats the problem.

>>24500264
What? Why are you getting so defensive and angry? If you're that OP then I'm sorry, you can't invade every thread that discusses chatgpt to try to defend your fragile ego. I'm literally just answering the OP topic for this thread and why it's a bad choice for /lit/izens to learn, with a specific and relevant example from /lit/ for why.

Here's an example of your failed argumentation in this thread:
>>24500201
>copilot
>Maybe try using something else you dumb fag.
>>24500210
>Only 12 year olds use fag AI's like copilot.
>>24500219
>Shows chatgpt does the same with the exact same prompt
>>24500223
>No argument present
>>24500236
>moves goalpost after being soundly and directly refuted

You need to unironically grow up. This is not a trial to defend your ego and you're shitting up every thread with your rage and hurt.

>>24500210
I think you should kys

>>24500272
Here comes the midwit cope again, you obviously can't process the humiliation of having your ego smashed to pieces.

You literally don't even have a grip on reality, trying to rewrite history when everyone can see you yourself conceded but crying about AI when it makes a mistake, this is pathetic, get some help buddy.

>>24500278
You are the perfect example for why the University of ChatGPT is retarded. No arguments, failed even in this thread, all rage. You're hopeless.

>>24500286
So desperate for a win aren't you. You're never going to learn anything if you always think you're right.

>>24500291
I'm not trying to get a win because it already happened. Why are you so hung up on winning? If you're that broken up about it, explain why you didn't reply to these three posts: https://warosu.org/lit/thread/24492095#p24494855
But you didn't, you can't, you won't. I'm not here trying to get a "win" on you. I answered OP's question with a relevant example and you came in shrieking despately about how you won like some insecure loser. The first thing you said was a desperate cry for help >>24500183 about how you "won" without any relevance to the OP topic.

I recognize the legitimacy of Peterson Academy(tm)

>>24493439 (OP)
No we don't

>>24493439 (OP)
I'm a codefag as well as a writer so I spend most of my working life interacting with these things. Since I work with code I can see empirically when it's right and when it's wrong. They are extraordinary learning tools if used intentionally, but they are also escape valves for laziness. 10/10 I prefer the perspective and experience of a competent/knowledgable/genius human over it, all things being equal.

From the literary perspective my view remains unchanged. I still don't think they stand up to a human being, not because they are lacking in the technical or mechanical fluency, but simply because there's no "there there". It has not true position, concern, care, perspective, or taste of anything. I love Flaubert because his humanity sings through his prose and I recognize it in myself. I only ever see a cheap imitation of that in these bots.

That said, I did create a "custom GPT" that is essentially a cyber-Flaubert. Occasionally it's an entertaining diversion to discuss literature with it, and its responses of "Ah!... that wistful feeling of so and so" whenever I bullshitted with it is consistently amusing. It even produced what I consider to be genuinely beautiful short stories and trenchant critiques of my writing by emulating Flaubert.

My thinking is: isn't a world where we can interact with these convincing emulations better than a world in which we can't?

All in all, I think these AI chatbots are a net positive. That said all the bull crap about these language models leading to "artificial superintelligence is a bad joke.

>>24500291
>You're never going to learn anything if you always think you're right.
after repeatedly brushing away other people's arguments, do you not realize how ironic this is?

AI and especially ChatGPT is cancer in its purest form

>>24502747
Just because someone has an argument that doesn't make them right.

>>24502747
>wah why isn't everyone listening to me
Shut up faggot

>>24502747
Works both ways, most christians just come here to proselytize while being in complete denial of reality.

>>24493439 (OP)
It used to be people were just stupid. Now we have AI being stupid for people.

>>24503030
In this case the anon was a retard using chatgpt unironically, using arguments with outright false information

>>24503076
Bro, are you still crying about that!? LMAO

>>24503082
What are you talking about?

>>24502747
>he samefags you 2-3 times and goes schizo immediately
Is it F. Gardner tries philosophy arc?

>>24502526
>I can see empirically when it's right and when it's wrong
One of the major problems with AI are the convincing hallucinations that can be peppered into a large text, and unless you are fact checking every single thing, you can't be expected to know what is empirically true or false, as that would imply that your knowledge is infinite. When a topic gets too advanced, obscure, or novel, and the more text is in its context window, the more likely for massive errors or that these small errors will build up to a critical mass. Even when there are minor errors, you can slowly build up your knowledge of the world based on a collection of convincing yet erroneous facts, and when a larger delusional error aligns with a constellation of these minor misunderstandings, you will be much more likely to think it's reasonable. From your perspective, the subtle ways that everything is just ever-so-slightly off line up perfectly with the greater delusion, so it seems fairly reasonable. When it comes to those really deluded facts, some of them are extremely difficult to verify, especially when they're novel topics.

If you really are in compsci then verifying AI is simpler in directly programming/engineering related topics, because you'll know if something is going wrong as soon as you type
>g++
But when it comes to literary critique, philosophy, obscure and novel topics, AI can be inaccurate in so many ways and the topics can be so difficult for AI the AI itself becomes kind of useless.

With that said, AI should never be used as a source of information or replacement for educating yourself through reading and contemplating on a text. It's best to use for quick pointers and looking up ideas and terms that you can verify yourself quickly, but NOT explanations of passages or texts. The best use for AI by far is general pointers, using it to get a normal person's perspective on something, as most of the training data is from midwit redditors, and especially fun stuff and banter like you have with your flaubert AI.

>>24493439 (OP)
I feel like a lot of people are not updating their impressions. o3 is kind of mindblowing. It does no longer produce vacuous verbiage like GPT-3.5 or even GPT-4. It's on a fundamentally different level.