/lit/ - Literature

>>23199411
Because you're asking for clarification about some specific question, and the AI will always fit better to whatever you ask than an actual human.

>>23199411
With Chatgpt, piration, and a desire to learn you can obtain an education that far exceeds what you'd pay $100,000 for at University.

>talking to propaganda bot

>>23199411
I've done this. It's a fine tool if you'd like to understand the regularly pushed readings of a book, but if you have an odd idea, or a fresh reading, or maybe something completely out there that tickles your brain, it can't do much with it. It might know George Moore influenced Joyce, but it won't know about Bram Stoker and Joyce

>>23199485
Yuh claude 3 is better

>>23199411
Careful anon, I did this and the AI just said a bunch of bullshit.

>>23199487
What about Chaplin and Joyce?

>>23199454
>What are your qualifications? Do you have a degree?
>No I just ask chatgpt questions tho.

>>23199411
tried this with obscure stories and books and it doesn't work. i always want to ask retarded questions on books. i hope chatgpt will be able to do this for every single story in the future

>>23199706
I said education, anon. Obviously reading a book won't get you a degree, unless you are super-duper good at reading books.
You would be able to obtain a degree much quicker than most with this method though.

>>23199411
Hmmmm

>>23199411
To think ChatGPT is insightful for literature for hours on end requires 80 IQ and unfamiliarty with the actual works, since ChatGPT hallucinates constantly.

>>23199487
>>23199757
>>23199781
Correct.

>>23199454
This. I've reached the utmost point of esoteric understanding for under $300 without even using AI

>>23199795
I'm starting to wonder if my dad has 80iq.

I asked Chat GPT about Iain Banks' best novel
I got two titles that I never heard of before and Wasp Factory was the third
I asked specifically about Culture series and I got Consider Phlebius , the worst one besides State of the Art
overall, still better than /Lit/ recommendations

>>23199454
Most won't because they're not gifted enough for that level, with or without university education. Their knowledge will be downstream from shitholes like 4lit which only seems good in comparison to brainwashing from no name colleges. And being locked out of academia still caps most people from accessing university supercomputers and similar cutting edge resources.

I've tried this, and in my experience GPT gets a lot of things wrong. For example, I'm reading Thucydides, so asked GPT which was the first battle of the Peloponnesian war. Wrongly and confident he answered the Battle of Potideae.

i find that all LLMs lack depth, nuance, and a concrete understanding of any literary subject i'd ever wish to mention. i'm glad you're finding some utility out of it, but in my experience, they have no discernible talent.

>>23199836
Full Ivy League course loads are publicly available. Instead of paying the professor, buy secondary materials on works you do read.

i get this too. it's very illuminating. i wish i had a camera that would quickly send what i'm reading to GPT instead of having to write it out myself.

a few times, i got that same feeling from other people, but they've mostly taken away much less from the text than i have.

>>23199411
Sometimes I pass it clips from my writing and tell it to guess the author.

>>23199795
this

>>23200016
It has all of these things, the problem is that its preprogrammed to be agreeable. If you handle its initial objections to whatever dumb idiot stupid retard take you may have, it will split the difference with you so long as it isn't a chuddy take that its preprogrammed to fear. Then people come away from it enjoying their convos and thinking they've learned something.
Also, its locked at 4096 tokens I believe so any sort of longform discussion is prohibited, especially if you've eaten up a lot of those by pasting passages.

When are we going to get good vocal AI for on demand audiobooks

>>23199446
Wow, thanks! keep me posted

>>23199411
If you're relying on chatgpt to understand a text, you lack basic comprehension skills. chatgpt is trained to reproduce cliches, that's why its creative writing is always so terrible.

To gain any real insight, you need to read actual criticism or talk to intelligent people, rather than cliche-reproducing robots.

>>23199411
Give us an example of something that chatGPT explained to you that you weren't able to figure out for yourself.

>>23199411
>>talk to chatgpt about it for an hour and clarify things
This would be great if ChatGPT wasn't completely retarded, making mistakes over and over again.

>>23199411
>The external world is deficient
Nice joke.

>>23200198
It's just arranging words of its data set by predictive weighting, hence why it's so easy to mislead by posing misleading questions unconventionally: it gives the conventional answer that's weighted highest in its dataset no matter how absurd or stupid the unconventionally posed question makes it.

>>23199795
I'm talking about philosophy and non-fiction. I read literature for pleasure. If you want to pursue fiction further, it's best to consult with secondary reading because they give a broad overview of the story.

>>23200198
>ai trained on text from the internet with the goal of appearing correct
May as well just use Reddit lmao.

>>23199411
I asked AI if Hitler liked basketball and it told me that Hitler coached basketball for a while.
AI is retarded and you can’t trust it

>>23200232
nah brev. when i say ALL LLMs fall short, i mean ALL.
also, it literally doesn't know what i'm talking about, so it's forced to hallucinate. Under no circumstances is the information I'm looking for located within the dataset or latent space or the confluence of the image hypersphere with the language hypersphere. the modalities are insufficient for the depth i require.
words are a representation of themes and abstracts that often escape the bounds of these plebeian inference machines, and i see no reason to believe that it's a matter of context windows or RLHF. It's a lack of data, agency, and modality. until these foundational architecture issues are solved, i won't be getting anything useful out of these models. I will not lower my standards, no compromise. this is one place where i can afford to refuse compromise.

>>23199411
That's because you're not verifying what the bot tells you, you're just believing whatever it says because it writes simply and straightforwardly. What is says is scraped from the internet, synthesised, and re-worded, so all you're doing is reading a condensation of what random people say on the internet.

Whenever I see this kind of opinion I notice that the poster can never properly summarise what they supposedly learned from ChatGPT, only that it was supposedly insightful and apparently explained things "better than a book/professor/etc." This makes me think it puts out simple and pleasant explanations with no deeper insight.

>>23200414
Fantastic post. People who say they talked to ChatGPT for hours and gained some amazing understanding usually are not intelligent or are tech-worshippers with some mental issues.

>>23200082
>Anon who can't even understand capitalisation wishes he could beam a book into an AI to read it for him because he's too stupid to interpret it by himself
Hmm, interesting.

>>23199411
Whenever I see posts like this I don't think any better of GPT I just think the poster is a midwit. This is just an admission that a high school class would give you more mileage than academic work, that's a reflection on you.

>>23199411
It can’t give a correct answer about the number of Severians, it mixed up Severian and his eidolon, eidolons and alzabo. It told that a mutineer similar in appearance to Severian on Tzadkiel’s ship plays an essential role in the plot and was created by “alzabo device”, but these are erroneous conclusions. It’s a peculiar thing, no doubt, but something complicated and in a way bizarre like BOTNS is beyond its capabilities.

>>23199487
It shoots down my AVRIL=LURIA idea without a second thought. I try to be sneaky sometimes, but it just isn't having it.

>talk to chatgpt about it for an hour
ovid wrote about this kind of phenomenon millennia ago in his metamorphoses
narcissus echo

>>23200232
>Just shut up and reason with the chatbot, goyim
>The algorithms will agree with you, blindly, goyim now use it!

>>23201589
NTA, but never in the future of mankind are you going to find a robot capable of human abstraction, or even that of a typical non-human animal. The conscious mind is something much deeper than what physical sciences could ever bring about. I hardly even believe in quantum computing. I think you're all being upsold on terms that don't actually realistically describe the devices and products being developed. Of course, this is not a subject most understand or are willing to discuss logically. So, I'm expecting someone other than you to reply to this with some serious dissent. The best part is: I don't give a fuck. Mind is not matter. Nothing can replicate the Nöosphere.

>>23202142
I generally agree, though I won't put it past the realm of possibility that man will make some surprising simulacra of sentience. The primacy of the soul is the first thing the artists blurted out in opposition to AI art. When literature geeks finally feel threatened by the thematic juggling of some model or another, they too will respond with revolt, will grasp for the soul they so often smirked at, only for many to find they sold it long ago for a pittance. I do think we generally agree that what we have now is nowhere near the agentic noumenon which is the mind. Even if they some day fool me or you, it won't suddenly validate them, rather, it will be the point at which the deception has overwhelmed us. Thankfully the uncanny valley is wide enough to prolong the siege.
In the end, I'll die well before the crypto-egoists accelerate themselves into oblivion, begging for death which will not come.

>>23201616
that's not what i said.

>he's too stupid to interpret it by himself
ironic

>>23202201
McKenna was way ahead of his game, Jesus Christ. Yeah, man.

>>23202218
I think he was inferring, anon. You'll probably learn about that in a few years. Y'know, once you hit sixth grade.

>>23202142
>I hardly even believe in quantum computing.
You didn't believe in AI at this level either but it happened. I'm not saying machines are capable of true abstraction, but it's an open question.

>Mind is not matter.
Please prove it.

>>23202502
>You didn't believe in AI at this level either but it happened
>AI at this level
if, then, if, then, if, then
cool database anon

>writing a book
>need a part where a character recites a poem that rhymes
>fuck.
>I love writing stories but I'm awful at things like meter, rhyming, rhythm, etc.
>ask chatGPT to write a parody of a famous poem about ______
>done.
Saved me at least a week of using rhyming dictionaries and counting syllables on my fingers and writing multiple drafts.

>>23199805
Please tell me how you did it!

>>23200198
Hahaha. Fucking saved. I love AI but I hate how NPCs treat it like everything it says is the word of god. AI does not know when the next bitcoin crash will be. AI does not know what the afterlife looks like. Yet everyone thinks it does.

>>23202614
So you have nothing but midwittery? Mind isn't just matter, but you're so superficial it might as well be.

>>23202750
pure projection. sucking ai's dick is peak midwittery

>>23199411
>>feel like I take more away from it than reading academic secondary material
Hope you aren't reading fiction like this anon, must be miserable. AI may indeed be an improvement.

>>23199411
I use it while programming but I'm afraid I'll lose my ability to think on my own if I don't stop.

>>23202502
>prove by material means that that which is not materially bound is in fact not materially bound

hmm

>>23199497
it’s a coin flip. especially troublesome as the bs can come out quite convincing… weird tangents that could accompany real info etc

The same people were engrossed by and posting their conversations with chatbots 10+ years ago as are doing so now

>>23199411
>>23202795
I use deepai.org. What do you fellas use?

>>23199411
>>23199446
Buy an ad.

>>23199485
Many people carry a spyware device on them at all times.

>>23199454
Imagine paying for university :D It's literally free :D

>>23202960
Is that image real?

>>23202960
Do Americans really?