maybe Robin fucking Hobb should try writing better characters and not emo slop

Here are the stories for the contest. If you have such good taste, can you tell which are AI and which are human? Can you tell which are slop and which are good?

https://mark---lawrence.blogspot.com/2025/08/so-is-ai-writing-any-good-part-2.html?m=1

>>24676176
3, 6, and 7 I had guessed were AI based on em dashes alone (I was wrong on 6), 2 was obviously written by a fart fetishist.

Tips for writers--use gamer words and you'll never be mistaken for AI.

Cool. Who the fuck is Robin Hobb?

>>24676264
One of the most acclaimed fantasy writers of the last 50 years.

>>24676274
>last 50 years
Irrelevant at best.

>>24676176
Hobb gets off on writing male characters being beaten, cucked, and shamed.

>>24676319
So? That should have been an advantage here, if anything. LLMs are muzzled.

>>24676182
>Robin fucking Hobb
Literally who?
>>24676250
Good
I hate niggers and kikes

I don't read anything published after the mid-20th century, so this has zero effect on my life.

>>24676632
You don't need to start a new thread for this

>>24676176
One more piece of proof that people whining about AI only hate it because they're told to hate it, and actually like AI generated work if they're not informed that it's AI

I posted my short story somewhere and got a couple comments saying it was AI. Such bullshit, makes me want to just quit

>>24676687
Correct. People are just scared of their replacement species.

>>24676176
>AI beats generic female fantasy author
Reading "Assassin's Apprentice" right now, it's a Netflix movie, 6.5/10 book equivalent.

>>24676699
Come on, anon. You know this is cope. How much longer until LLMs are better than every human author who ever lived?

File: holowhistle.jpg (33 KB, 453x576)

33 KB JPG

It is a blessing. Now that our digital puppets can successfully replicate “tumblr-speak”, “reddit-speak”, “HR-breakup-type-prose”, “fan-fiction”, and “Guardian short story competition winning entry” type-speak — living writers will now be propelled (forced) towards increasingly inventive usages of language and other literary devices. The wheel of prose will turn. It will enter a post-AI age, where originality will be necessitated, and language will be defined by its overwhelming humanity!!

>>24676176
>"readers"

File: 1551443066959.gif (607 KB, 300x169)

607 KB GIF

>NOOOO I'LL HAVE TO BE CREATIVE FOR ONCE AND STOP RIDING ON THE FANTASY DROUGHT COATTAILS
I voted for all humans, even the ones I knew were obviously AI

>Robin Hobb
>Rub a knob
Based on a true story.

>>24676640
I actually didn't, there's a copycat anon!

>>24676176
>Robin Hobb
Landfill fantasy

>>24676702
LLMs will always struggle with writing anything truly unique or innovative. That's just how they're designed. They're already better than most authors, if used as a tool, fed a detailed outline and other key supporting documents, and heavily line-edited. But they can't effectively write something off the wall like Palahinuk or Vonnegut.

It will likely result in a similar response, long term, as the art world had to photography.

>>24676894
>LLMs will always struggle with writing anything truly unique or innovative
so will humans. Nothing new under the sun

>>24676274
>fantasy
Not a writer.

>>24676908
Cope. It can do "literary" fiction just as well, but no one has bothered to test that because no one cares about litfic.

>>24676911
Ok, prove it.

File: 1000023658.jpg (31 KB, 400x400)

31 KB JPG

>>24676927
Picrel is AI

>>24676927
Use GPT-5 yourself, dummy.

>>24676949
>>24676927
It cant do shit. Gpt 5 is horrific.

>>24676894
This guy gets it. It can iterate in the style shared with it and improve upon it, but it cant create, and when it does create, it is always mid.

>>24676176
>Robin Hobb
nigga who?

>>24676953
>>24676953
>it can't do shit

On a windy morning, Lin found a violin on his doorstep. The case was black, zipper broken, blue twine tied hard around the handle. He carried it inside to his bench and opened it.

The varnish was amber. A long split ran from the treble f-hole toward the soundpost. Bad news. A soundpost crack. The bridge leaned; the strings were tired; dust lay in a pale drift inside. The instrument smelled of resin and old sweat.

Pinned to the lining was a square of paper.

For Lin: Rowan’s. Please help if you can. His daughter wants it to sing at the memorial. –T.

Anyone around Lydiard Street knew Rowan: the baker’s-door busker with old tunes and a gentle hat for coins. He’d died last week. Facebook had filled with shock and casseroles.

Lin heated the kettle and set a jar of hide glue in a water bath. He called the number.

“Tess,” the voice said. “His sister.”

“It’s a soundpost crack,” Lin said. “I’ll have to take the top off. It’s delicate. The memorial—when?”

“Sunday,” she said. “If it’s ready, it’s ready. If not, we don’t make it worse by forcing it.”

“And the daughter?”

“Caro. Eleven. It was her idea to leave it with you. She says wood listens.”

“Sometimes,” Lin said, and meant it more than he intended.

He took off strings and bridge, warmed a palette knife over steam, and worked it into the seam. Hide glue smells like a shed on a cold morning. The top lifted slowly. Inside: a low ridge for the bass bar, tool marks from some factory, an old corner repair.

At the crack’s center the wood was crushed into a shallow dish. It needed a soundpost patch: carve a hollow, inlay new spruce, fit it thin and tight. He traced the dark rosin stains and saw, faint in graphite, four notes with a tie over the first two. Next to them: C for Caro.

He put the plate down and rested his hand on the spruce. He would leave the pencil there.

He shaped the patch all afternoon, shaving and offering until the grain lined up. He glued it, clamped it with shaped cauls, and let it take. While the glue set, he cleaned the fingerboard, polished the pegs, and carved a new bridge to a height an eleven-year-old could manage. He cut the heart and tapped for the ring he wanted.

Near dusk, Tess and Caro came. Caro hovered in the doorway in a jumper with one sleeve pushed up and the other down.

“You took it apart,” she said, stepping in.

“I needed to see the hurt,” he said.

>>24676966
She stood on toes to peer into the open body. Inside is mostly braces and dust, and yet people always look.

“There were four notes written in pencil,” Lin said after a moment. “With your initial. I’ll seal them under shellac so they stay.”

Caro’s face flickered through a set of thoughts and settled. “He wrote on everything,” she said. “Receipts. A parking fine once. He said memory needed somewhere to wait.”

He glued the top back on that night. In the morning he set the post, tuned slowly, listened. A new E sang thin and hopeful. He drew the bow across the D. The voice came rough, then cleared. He played the four penciled notes. They landed under his hand like something familiar.

The memorial gathered by Lake Wendouree where the reeds hold the shore. Thermoses, folding chairs, a table with clipped photos. A speaker insisting it was paired to the wrong phone. People spoke in low voices that didn’t quite fit the open air.

Caro carried the case by the blue twine. She set the violin under her jaw. Her breath trembled once, then evened.

“You don’t have to be what they want,” Lin said quietly. “Just be as honest as you can stand.”

She nodded and began. The first note wobbled. The second stood up. The next two stepped forward. She found the penciled four and laid them plainly, the way you put important things on a table. It didn’t sound like Rowan. It sounded like a kid who had listened closely and would keep listening. The patch held. The bridge did its work. The soundpost stood where it should and passed the weight on.

No one clapped. People who are swallowing tears don’t. Tess touched her niece’s cheek. Someone gave Lin tea in a paper cup. A man asked him about an old guitar. A child asked if violins were hollow because music needed somewhere to stand. Lin said yes.

When the crowd thinned, Caro came back and held out the violin for him to look at, not to take.

“Do you think he wrote more in there?” she asked.

“I think he wrote what he needed when he needed it,” Lin said. “You will too. Maybe not on wood.”

“Mum says I should go back to lessons,” she said. “Scales until my fingers feel like they’re wearing gloves.”

>>24676967
“Scales teach the map,” he said. “But no one lives on a map. If you get lost, play something you remember from the kitchen.”

She looked at the lake. A white feather was caught in the reeds. “When you glued it,” she said, “did the violin know?”

He could have explained capillaries and glue. He said, “It knew how it should vibrate. It was waiting for someone to give it that shape again.”

She stored that away and went to find her aunt.

Back in the workshop, the room held the day’s smell. Lin put his tools in their places. In the drawer with the spare bridges sat his mother’s metronome. A distant train passed. The floor ticked. The metronome clicked once—its little ghost of time.

He stood with his hands on the bench and thought of the pencil marks under the plate: four quiet notes, sealed in, waiting. The instrument was gone, but a thin thread of its voice seemed to hang in the air, not a tune so much as proof that air can remember. He switched off the light. Outside, leaves rubbed together. The room kept a small noise of its own and then went still.

>>24676970
this shit is like broth with far too little water. its gormless rigaramole. its like a bookclub fucked a coffee shop. its like SSRIs were born in seattle. its completely hollow. words are containers.

>>24676966
Yes, it does well in the <1500 words space. Nothing exciting - it won't play with language or use novel framework - but very workable.

Try to expand it to 10k words, much less 100k, and you see the cracks. Or have it write a dozen different 1500 word pieces, and you'll see the similarities in what the model outputs.

It's impressive what it can do. But it can't write a coherent, full length story without heavy editing and a ton of time spent on structuring the story first.

>>24676274
>Societal prejudice against the ability causes Fitz to experience persecution and shame, and he leads a closeted life as a Wit user, which scholars see as an allegory for queerness. Hobb also explores queer themes through the Fool, the gender-fluid court jester, and his dynamic with Fitz.

I mean it's a woman so I should have known it would be bad.

>>24676687
But what is the point of reading AI generated regurgitated slop based on the blended ideas and works of actual people? It’s almost like when you know something is literally soulless it feels like a poor use of your time

>>24676176
Let's see it beat Gene Wolfe.

>Soulless slop from authors are indistinguishable from soulless slop from AI
Never heard of Robin Hobb before. Who the fuck is Robin Hob? I never want to even find out what the hell a person called Rober Hob is writing. Why did you think this was appropriate, I don’t care about Rober Cob? What the fuck is a Robert Cobb? BOB COBB? BOB COBB! BOB COBB COBB COBBBB

>>24676176
this says more about readers than it does about ai

>>24676970
Was I supossed to be impressed? Technically proficient, but diagnostic and impassioned. Reading this was like watching a caterpillar inch along the dashboard of your car; neat, but ultimately aimless. The entire time I was reading it I kept wondering "what's the point". This is the expected quality of a below average MFA, one with technique and rigor but a misaligned goal. It would only impress the dreariest of judges. If I wrote that? No one would care. If a disabled black jewish woman wrote it about her experiences as a child? Probably award winning, but still not impressive.

>>24677310
This. Im sure the average normie couldn't make out AI writing. We can also assume the people they tested it on usually read booktok type shit as most readers sadly do. If it's about copying that style than im sure the AI could do it better than Hobb.

>>24676194
when in doubt, always accuse AI. It's better to shoot ten innocent guys than let one robot walk

>>24676966
it's actually worse than I would have expected
one can imagine the red pen markings almost immediately
>found a violin...and opened it
would be derided for its lack of clarity. there's no reason not to say 'The case [that housed it] was black...' instead, since as-is we are pointlessly made to infer that the violin was in the case from the fact that he carried 'it' inside and that the next paragraph describes the state of the instrument itself
>Lin heated the kettle and set a jar of hide glue in a water bath
again we have a senseless lack of clarity: he took the violin case to a bench and opened it, but when did he get up to put 'the' kettle on and prep a water bath for the hide glue? where is he now, and where are these items?
>He called the number.
where did he get 'the' number? if it was on the square of paper pinned to the lining, why was this not indicated?
etc., etc.
that being said, a decently-capable writer can quickly identify and correct these flaws
it's a force multiplier for slop production

>>24677493
Honestly if we’re going off the average reader’s take I’m sure AI would win out over Poe, Lovecraft, Maupassant, Chekhov and others just based on the fact it’s writing in easily understandable, simple, modern prose regardless of content

>>24676704
>hopecore posting
>in 2000+20+5
ngmi

the average reader in 2025 likes shit like brandon sanderson and generic romantasy slop, of course they're going to prefer the ai short stories

If a story has logical progression with fluid prose and a clear main point organically derived from the events of the story, it must be AI. Humans only wrote meaningless lolrandom edgy trash these days

>>24676264
>>24676629
>>24676955

I think Robin is the one who just did that school shooting?

AI is way too good at writing. No way around it, it's bleak. The implications. The future now before us. Ayy Eye bodes ill, spells doom, is bad.

>>24676176
In my teens I read all her books twice and yet I don’t remember a single thing

i dont see how ai can be all that relevant unless it is / becomes a consumer / individual consumers

>>24677680
All literature but especially this fantasy slop has been deliberately cut down to be as cookie cutter as possible so that the illiterate general public don't have to engage with the words on the paper and just get a movie script injected into their eyes, and somehow they've decided to be upset about the logical end of their pathway.

>>24676194
all of them were meh. but I couldn't tell which one was the real writer. it's so over. I suspect story 7 could be real because It was cute and unique but who knows.

>>24678892
nah i just checked the answers and 7 was AI, killing myself

>>24678899
>>24678892
it's over. art has officially died.

>>24676194
Wtf is going on with story #2

>>24676176
>we're collectively awful at identifying AI writing
Was the task to identify AI writing, or identify which writing they liked more? The article isn't clear.

>>24677310
I was about to say this. It doesn't say *nothing* about AI, but it says much more about the people being tasked. We already had a half dozen threads in the last two months about that study of English majors from 2015 who struggled to understand the opening paragraphs of Dickens's Bleak House even when allowed to use their phones to look up things. It's not a shock to hear that some general readers today can't distinguish between AI and a mid author.

Love that bitch. Read all 9 books in college as a security guard and would go on rounds whenever I teared up.

>>24676194
Out of the 8 stories I only guessed 3 right and I was very certain of my anwsers, not so much now.
The only one I was halfway guessing right was #6 as the concept of being a tennant in your body was interesting, but the inital part was so bad it made me vote AI.

>>24676194
gave up halfway because all of those sucked

>>24677279
That applies to the vast majority of human-written works as well.

>>24676176
>Robin Hobb
Who?

>>24676250
are em dashes an ai thing now? I use them all the time

>>24679901
it was trained on the midwit phd style guide

>>24679860
#deep bro

>>24676176
have they tried against good writer rather than genreslop written by a feminist?

>>24676274
>One of the most acclaimed fantasy writers of the last 50 years.
Really? I read a fuckton of fantasy, including modern, and I can't recall that name for shit. How many Hugo/Nebula/Locus awards this one of the most acclaimed writers has?

>>24679923
It's not deep or pretending to be. It's just the way it is.
LLMs are getting better and better, and the portion of artists that are still better than it is getting smaller and smaller.
For most authors the day where they get surpassed by AI (with some amount of curation) has already come, so why would anyone who's not among the genius elites take advantage of it and get something superior to what they would otherwise have? There's a full spectrum of AI uses that go from using it for basic research and "advice", to letting it make the whole thing from scratch. For each artist there's a sweet spot where the pros outweigh the cons, and the end result ends up being better, while taking less time to create.
If they care about the artistry they can have fun writing 100% human works in their free time (of which they can have a ton of by outsourcing to ChatGPT lol), but if it's their job to deliver books that people are going to enjoy, it would be retarded to still do everything manually; especially in a world where everyone is competing against an increasingly "AI-augmented" competition.

The whole "humans are still better than AI" doesn't really make sense anymore outside of a few exceptions.

>>24679963
>The whole "humans are still better than AI" doesn't really make sense anymore outside of a few exceptions.
AI (hell, even pre-AI algorithms) always was and always will be better than humans at manufacturing slop.

The problem is not this disparity. The problem is the very existence of the slop market.

I read all the flash fiction pieces and they were all garbage. I rated them all one star or two stars. I think there was one three star, and I only rated it that because it held my interest to read it completely.

The ones that were horribly cliche and used tired old expressions I judged as being human. Also the ones that were incredibly melodramatic I judged as being human. The ones that were artsy but clunky and had sentences that didn't flow together I judged as being AI. I ignored all em dashes, because I bet that's a red herring.

I won't be surprised if my judgement between which was AI and which was human is wrong. They're all garbage and pretty much interchangeable in terms of quality.

>>24679978
The slop market can only ever be a constant, simply because among writers not everyone is great, and among the audience, not everyone wants the greats (for a variety of reasons).
In the continuum between the sloppiest slop and the greatest masterpiece, AI's output quality is well past the halfway mark, and wherever your personal threshold is, AI can and will help a lot of people who would normally be below it to clear it.

Plus, as it continues to inch closer and closer to the level of the greats, less and less slop will be released because the bar will be so high by everyone using it at least to some extent to improve their works.

>>24676194
I stopped reading at
>This time I've convinced a number of very accomplished authors, who you will have heard of to provide the human-authored contributions.
I can't take the author seriously, who cares whether he's able to tell human and AI writing apart when he can't write a phrase himself, let alone a story.

>>24680000
You don't need to be great in order to not be slop.

>>24680021
Did you even read my post?
I specifically talked about it as a spectrum, with a subjective cutoff point below which one may consider something slop, and above which they may consider it non-slop.
The "greats" are merely the top end of the spectrum, not merely what's above the non-slop threshold.

>>24676176
Robin hobb isn't a short story writer

>>24680041
What’s your definition of the “greats”? Because it clearly doesn’t involve originality or inventiveness of language

>>24677279
It's fascinating how this hasn't been an issue in the art world since the debate on readymades got settled, what, a century ago? But when it comes to the cultural industries, suddenly there's an army of people defending the soulfulness of the worst and most derivative garbage imaginable. Every online discussion about AI art devolves into arguing if it is or is not as soulful as fan fiction, comic books, cartoons, or other similar generic shit.

>>24680594
You're assuming that the current state of LLMs will be their state forever, which is absurd when they went from barely doing the keyboard suggestion thing to being more creative than most people (I know it's not a high bar, but it's still an unfathomable improvement) in just a few years.
The way these models are finetuned and augmented with all sorts of non-ML software components, their capabilities are radically improving at an extremely rapid pace, even despite the upgrades in compute/hardware slowing down.
The prediction that the quality of its output will improve drastically, and that companies will make specialized models for every/most applications (including /lit/) that are preprogrammed by humans with the characteristics most conducive to a high quality output (see Claude and the other programming ones), is the conservative one, and it's been proven true time and time again while you retards kept saying "ok but it can't do X right now so it will never do X", then (after it's able to do X) "ok but it can't do Y right now so it will never do Y", etc.
Utterly retarded.
Like seeing a fleet of warships getting close to your shore and gradually going
>nah there's nothing to worry about, they're too far away
>nah there's nothing to worry about they're just on the beach, they're not doing anything
>nah there's nothing to worry about, they're just having a scuffle at the city doors, they'll never get in
>nah there's nothing to worry about, they just wanted to kill the king, they're not going to do anything to us
>etc...

>>24680041
>Did you even read my post?
Yes, and you are an idiot.

"Greatness" and "not being slop" are not even on the same axis.

>>24680664
I'm approximating "quality" to one variable for the sake of the discussion you retarded autist.
Of course there's an infinite number of factors at play and that "great" works can be great for a variety of completely different reasons, but that doesn't change the point I'm making, which is that every characteristic that makes the greats great, LLMs have been improving in at an unbelievable pace.
Everything they used to make was sub-slop trash, and now it's hard to distinguish from the work of professionals (objectively less slop than before).
If you're too dumb to extrapolate the most obvious trends, that's on you.

>>24680594
Not that anon, but there are plenty of great novels lacking in originality and inventiveness of language. But even then, the idea that LLMs can only spit out collages of their training data is likely wrong. Or more precisely, it's wrong to assume that the training data isn't enough for them to do things we may consider original and inventive. I used to think otherwise, as it would seem common sense that a LLMs could only give back what was put in, but I think Sutskever is right that even current models are capable of extrapolating to a far, far bigger degree we expect.

>>24680678
>which is that every characteristic that makes the greats great, LLMs have been improving in at an unbelievable pace
No.

There are great slopwriters, and LLMs can be better than those right now. In fact, as LLMs improve, they actively lose their capacity to surpass the greatest slopwriters, as they sufficiently uniquely braindead data loses relevance in the ever-growing training datasets.

There are entirely mediocre and even poor writers who are not and never will be slop. Some of them can be replaced by LLMs in their entirety right now. Some of them can never be replaced by LLMs, because they can't even be replaced by another human, or anything else.

You are dumb for trying to reduce the concept of /lit/ to a two-dimensional graph that you can extrapolate like a 92 IQ pajeet mutt.

>>24680662
It seems like we’re conflating greatness and commercial viability. Companies may want an AI that creates the next Court Of Thorns and Roses or Harry Potter, or can churn out Song of Ice and Fire or Colleen Hoover or Danielle Steele ad infinitum but what incentive is there for an AI that writes the next Moby Dick (which sold like shit in Melville’s lifetime)? And, again, this assumes people will knowingly, willingly pay to read AI slop.

>>24680693
You have no idea how LLMs work and what their emergent abilities are.

>You are dumb for trying to reduce the concept of /lit/ to a two-dimensional graph that you can extrapolate like a 92 IQ pajeet mutt.
Imagine getting sidetracked this much by the form of an argument when its substance was literally spelled out for you as something different than what you misunderstood and and got fixated on, then doubling down despite everything.
I'll give you the benefit of the doubt and assume that you just have trouble admitting to being wrong, and that you're not actually this mentally challenged.

>>24680697
>It seems like we’re conflating greatness and commercial viability
I'm not.
What makes you think that the things that make something great can't be emulated by an LLM with the right tuning, prompts, etc?

Mind you I'm not saying that we will definitely have an AI that can write unparalleled masterpieces. Just that it's a possibility that we can't exclude a priori like you're doing now, and that judging by how much it's been getting close to that point, it's safe to predict that it will get much closer.

File: IMG_9515.jpg (367 KB, 1110x610)

367 KB JPG

I voted exactly like the average person

>>24680738
2 more weeks

>>24680948
You retards kept saying this about everything that you thought would never be possible but later became reality.
What makes it doubly retarded is that this isn't a prediction of a discrete event (like for example the coming of an AGI), but the simple continuation of what's been going on.
If anything I should be saying >le two weeks meme about your dumbass prediction that improvement will for some reason stop one of these days before AIs gain whichever ability your goalpost is currently at.

>>24680706
Nah, you're retarded.

>>24676687
I don't hate AI because it's bad, I hate it because it's AI. If I find out I've been fooled into thinking that something made by an AI was human-made, I'm not going to come away from it marveling at the AI's abilities, I'm going to be annoyed that I wasn't actually engaging with something made by a human being.

>>24681036
Whats actually going to happen is that once the internet is flooded with AI written works, AI will then scrape these AI works, creating a cannibalistic loop.

The end result is that the internet will get infinitely shittier, but people will be able to recognize AI writing. This process has already happened for AI art.

File: 2C9BEACB-D51A-40F2-933E-8(...).jpg (49 KB, 1280x536)

49 KB JPG

How did two pre-trans BDSM nerds predict this?

>>24679901
Yes they’re mainly an ai thing because it can’t comprehend grammar. You merely use them because you’re a pretentious twat who wants to feel superior.

>>24676194
Guessed correctly everything except 1st. Also, it was the only one I bothered to read till the end.

You can actually see somewhat distinct styles that human authors are going for. Could've made correctly first one too, but I just haven't read many flash stories made by AI in that context till that time.
My guess is, if you actually make people complete a bunch of those tests and allow them to see and ponder on the results, many will start to distinguish AI and human authors like 80% more easily. It's just a "skill" that isn't much needed or trained for now.

>>24680738
>we can't exclude a priori like you're doing now
I'm not.
I asked a question you didn't answer. I'm not saying AI can't write things people will enjoy. It's probably able to write things that are pretty good/ replicate greatness based on various prompts in a vacuum. I asked what incentive there is for a company to create, publish, and distribute a work via LLM that is not commercially viable, or, as in the case of Moby Dick with its long descriptions of types of whales, etc, that isn't interesting to people for another 50-70 years?

>how much closer it's been getting
Do you have other examples or just the present one where it's nominally beating Robin Hobb in a reader based short fiction contest?

>>24676984
>It's impressive what it can do. But it can't write a coherent, full length story without heavy editing and a ton of time spent on structuring the story first.
>
I played with a free one online. Given a specific setting and told a "tone" or "style", I admit it was a novelty to read the little story it would generate. I wondered about fetish sex, and I prompted it that a 30 year old girl would give her boyfriend a hard spanking with a riding crop. It was actually interesting. The girl made a cute/hot speech and scolded and threatened him. She only actually hit him with the riding crop once really hard, on his thigh. I wasn't expecting that. I prompted it that the crazy serial killer had a helpless victim in a dark basement. It did a credible genre slop job of it.
>
What I can't get out of it? Meaningful and purposeful dialog. Dialog tends to be to the point and sparse. I had it do a cop interviewing criminal to a confession? Sparsest dialog I could imagine for it. This should be dialog heavy! Its an interview.
So to me simple proof author isn't an AI is simple. Look for a dialog chapter. banter isn't the easiest thing for a new writer to do, and it reads natural. As a writer if I used AI I would have to match my writing to the tone and style to what the AI produced. I seem to have to babysit and refine prompts to shit out 1k words at a time. Choose one and edit it. I could have written 1k words in that time.
>
I would like to try prompting for fun and getting inspired by one good on it shit out. Expand it into a longer work, with no further AI interaction.
>
Another notable failure was I kept asking for a simple SF scene. A guy is on the run from the law, so he steals a spaceship and escapes after being chased and shot at by other spaceships. AI literally had no idea what to do with that request. It floundered and produced meaningless crap over and over again.

>>24676970
>vibrate
>ghost
>thread
>remember
shit stinks like LLM slop. unreadable

>>24676274
reading fantasy is like listening to metal

>>24681503
You think they're not aware of this and will just let it happen?
You don't think that especially a /lit/ model wouldn't have its weights and biases favor confirmed quality works and would just be allowed to ruin itself with LinkedIn posts and whatnot?

>This process has already happened for AI art.
And yet people are already making great-looking, snd sometimes almost completely photorealistic, images with them, with as much "artistic quality" as their prompts have.

>>24681632
>I asked what incentive there is for a company to create, publish, and distribute a work via LLM that is not commercially viable, or, as in the case of Moby Dick with its long descriptions of types of whales, etc, that isn't interesting to people for another 50-70 years?
There are many publishers publishing all sorts of niche stuff, not to mention people releasing the books on the Internet (through Patreon or directly on their social media) if there's really no commercial interest (unlikely, since the Internet allows you to find niche audiences and become successful without having the mass appeal that would've been necessary for even the smallest amount of commercial success in the time of your examples).
Besides, people have already released AI-generated books. The Internet bubble of ideologically-driven AI hate doesn't reflect the real world, where most people don't care.
If someone were to actually make a great work with AI, only a small portion of the target audience will think it's icky and avoid it.
And that's IF they're even aware it's AI-generated at all. A sufficiently advanced model capable of creating something very good, is likely to be indistinguishable from human-made works (especially if the "author" rewrites its output to hide its tells).
The AI variable is a non-factor for financial viability, and if you're asking specifically about companies (kind of a pointless question considering that anyone can use these models, not just companies), other than the aforementioned niches, any AI and AI-adjacent company dealing with literature would love to publish high-quality AI-generated works as proof of the capabilities of their product.
Open-AI, Google, etc already spend billions of commercially non-viable projects for this reason (or just for research).
Once the output quality is sufficiently high, what's stopping Silicon Valley from opening or making deals with major publishers and bookstores to sell their AI-generated works, even at a loss, just as advertisement for their models?

>Do you have other examples
The entire field of LLMs itself.
These individual cases are anecdotal at best (you can google things like it passing the bar exam 2 years ago, that paper about its poetry being indistinguishable from human poetry, beating Turing tests, etc)
Every measurable metric shows it's getting better and better at an insane pace.
What is your point? That it stopped improving? Do you have anything that even suggests that?

>>24676176
It was probably 90% women. Women are stupid.
Women made 50 Shades of Grey a huge bestseller, and it is considered by critics as one of the dumbest books ever written.
No surprise women can't tell AI from human writing--when it's obvious to someone who understands writing.

>>24676687
No, it's a statement on modern authors being so shit and high on their own genderqueer farts that modern readers find AI slop to be a better reflection of the human experience.

>>24676194
Haha. Wow. I was completely wrong. All these stories were trash, but how awful a writer must you be if everyone thinks your writing is AI?

The AI pieces were more thematically consistent and less cringe. The human pieces were cliche and trite and trying too hard to impress. I had it completely backwards. I thought it was the AI trying to be fancy and the humans were being consistent. Guess I overestimated human writers.

>>24676194
he should have done this with opus instead of chatjeetpt

>>24676194
The first three human stories were my three one stars. I guess AI is less likely to be offensively bad. However I also knew they were human

Why does it matter if someone can tell AI "writing" apart from real writing? The AI "writing" is still worthless and the human writing is still superior.
If you put a line of sugar and a line of cocaine on the table and told me to pick the correct one to put in my coffee, I might fuck up and choose the cocaine, but that doesn't make the cocaine just as good at sweetening my coffee as the sugar.
The only reason you would think otherwise is if you are a relativist who thinks that something being art, or being worthwhile art, depends on if it is recognized by the average person as such. This is completely false.

>>24682968
>The human writing is still superior
Part of the problem we have is that our last few generations of humans have been deliberately educated into dysgenic habits that keep them from being any sort of competition to major corporations. They sabotage their own products and ideas with things like LGBT, browns, ugliness, and other things humanity finds naturally repulsive. People keep telling them to go back to drawing, writing about, and writing for, beautiful white people of good standing and especially to appeal to young white men, who are the ones who build civilization, but they keep refusing, and here we are. If people won't make the content, we'll just make machines that'll make that content for us.

Humanity can go back to making worthwhile art whenever they're ready. Simple acknowledge that feminism was a mistake, LGBT is an irrelevant minority and nobody likes browns, not even other browns. Simple as. Jesus fucking christ, its not that complicated.

>>24676704
test--

>>24676319
And that's a good thing.

the """creatives""" will never stop coping untill they are fully replaced and forced to sell their 2k macbooks pro for rice

>>24683598
The creatives were coping from the start. Billions of human beings have been born, lived and died their entire lives without contributing a single noteworthy thing to humanity other than hard labor and extracted wealth. The art, architecture and writing that has been passed down to us over the centuries has largely been the product of the rich, powerful and well-educated, and now we have an entire class of unwashed, miseducated subhumans trying to pass themselves off as being equal to the best that humanity has ever minted when they're not even close.

The fundamental issue is simple - most people are not, nor will they ever be, equal to the top 2% in anything they do. No amount of retraining, reinventing yourself, or even hard luck will get you there - you would simply have to accept being a peasant, and allow the sum total of your life's work to be extracted along with that of millions of other laborers, so that it can be added to the handful of people on the planet that are better than the rest of us.

That's how the great works of literature and media have always been made, and it was always a cope that people are equal and interchangable. They're not.

>>24682968
>human writing is still superior
Not for these flash fiction pieces. The AI writing was much better. But that doesnt necessarily make AI superior, but the AI is clearly superior to these particular fantasy writers. They were so bad they couldnt make 800 words palatable. These hacks should be sweating bullets.

>>24682352
>Every measurable metric shows it's getting better and better at an insane pace

No, the opposite. You're another gullible fool who has fallen for AI marketing bullshit. It is not good now and is not going to get much better.

>>24683598
the real "creatives" already saw this coming.
https://m.youtube.com/watch?v=OHQRo3Uz_VQ

>>24684003
big latte no sugar

File: 1741118909175617.jpg (60 KB, 640x480)

60 KB JPG

>>24680697
Reading and literature in general are in a dark age and will probably never recover from the one-two punch of smartphone and tablet availability. It is simply a dead clade walking. In a couple decades, bookstores will be as rare as vinyl shops in the 2000s. Physical books and reading will only continue to exist as a performative marker of status. The women who keep many publishing houses alive today through voracious demand for romance novels will switch to AI boyfriends as soon as possible. Online places like this will be the only places to find good new shit to read.

>>24684003
>Good
Immigrant labor isn't 'good' either, but it dominates the market. Before that, slave labor wasn't as 'good' as artisan labor, but it too dominated the market. Beasts of burden were often not as 'good' as human labor, yet they dominated the market because they could be paid in food rather than in currency, until sharecropping became the norm rather than the exception, post-Enlightenment.

'Good' products are not what dominate the market. Cheap labor, and cheap and convenient products, are the order of the day. What labor is cheaper than free, because we all acknowledge that machines have no rights and aren't people? It'll be the 'good old days' for manufacturing all over again.

>>24684003
>No, the opposite
It's been getting worse?
Lmao

>You're another gullible fool who has fallen for AI marketing bullshit.
/x/-tier delusions.
You can literally use them. One year ago it was noticeably worse, and two years ago it was retarded compared to now.
People are now running local models on their laptops that outperform the flagships that a few years ago could only run on million dollar clusters.
What other technology makes generational leaps every couple of years?

>it's not good now
That's your opinion based on your subjective definition of "good".
Opinion that is also irrelevant to the point you (for some inexplicable reason) think is a counterargument to me saying it's improving.
Whether it's reached your personal cutoff for "good" or not, it's purely factual that it's improving at a rapid pace.

>and is not going to get much better.
Baseless assumption that conveniently happens to align with what you so angrily want to believe.

What an utterly worthless and fallacious post. Please learn to think before you post again.

>>24676194
I got either 6/8 or 7/8 right, forgot what I had the first one. To a certain extent you could infer whether the stories were real or not:
>#1
Has a theme and a sense of subtlety. AI can't foreshadow (yet) so that was the biggest hint, although the first paragraph was so generic that I was still concerned.
>#2
Also makes a subtle point -- it doesn't spell it out that the angel is corrupting itself by conversing with demons. An AI would have ended it just that way.
>#3
Depthless. Just generic action. Ending line makes no sense (why would a priest dissuade people from praying?)
>#4
This one I confused for a real person because of the three lines that start with "The", as in my head only a human could make such an ugly mistake. But still, the story is surface level and the twist has no power, which should have been enough for me to know it was AI.
>#5
Also powerless, and without any depth beyond "X happened". Though some of the imagery is nice.
>#6
This one I didn't like but it does clear itself by virtue of its structure. AI likes to keep everything nice and neat, yet this story is messy in how it digresses and starts rambling about property law and shit. Also just look at it, the division of paragraphs.
>#7
Immediately marked it as AI when it opened with "It was a damp Tuesday afternoon." Literally one degree of separation from "It was a dark and stormy night", if it wasn't AI then the author would have had to answer to me.
>#8
Doesn't reveal itself to be a demon story until the end, and even then it's still open to interpretation. An AI would have wrote flat out "the demon laughed".

Of course, these are established authors that the AI is going up against, with years upon years of experience to give them distinguishable styles. Against amateur writers I imagine the AI would be nearly impossible to smoke out.

>>24679901
Everyone fucking says they use em dashes all the time but I never see anyone writing with em dashes, especially online. You'd think 4chan and reddit would be half em dashes with how many people say they're always using em dashes but no one fucking does. It such a weird thing for everyone to lie about I really don't get it.

>>24676194
This test is hard because all of the stories suck. How can each of them just be so boring at the start?

>>24684142
I has similar thoughts in most cases, but I'll give my own impressions.

#1 Far too thematically purposeful, deliberately written, coherent and full of thought out idiosyncrasy to be AI. I would be genuinely shocked had this one turned out to be computer written, as it contains none of the faults that I see with actual AI writing and and uses the simple base scenario to good effect. I'm surprised how the poll results lean so heavily towards picking this as AI, it's the most obviously human writing of the lot and should stand out to anyone familiar with how AI works tend to go, as something beyond that.

#2 In contrast, my basic thoughts are that this is just way too *bad* to be AI written. It is the product of an ignorant, diseased mind lobotomized by generic fiction and conformity to the values of the modern liberal zeitgeist, I just couldn't expect an AI to create something this stupid unless very carefully prompted to.

#3 I was split about this one. I would have picked AI for this, but it had enough specific game fantasy cliches, like expecting weapons to be 'blessed', that it smacked of a bad human fantasy writer. The priest is presumably saying "prayer is not enough, these demons can only by beaten with physical weapons", which feels D&D enough.

#4 The first very obviously AI one. The way the demon simply breaks his word without doing anything clever feels like such an AI thing to write (demons are typically bound by promises), it's the kind of thing the AI would do because its just trying to make the story end a certain way even if this makes the whole thing pointless, and lame. All AI 'horror stories' tend to play out like this, in contrast to how planned #1 felt.

#5 This one very much surprised me as being AI. The picture it paints is quite vivid and there's a sense that an author was trying to suggest something specific like "what if a demon magically let you wish Drumpf away". It makes some sense on reflection though as the whole thing is vague overall and the impression of an ultimate point seems to be faked, but I was definitely caught out by this one.

#6 Definitely leaned towards human, though I wouldn't put it beyond an AI to write with this style.

#7 Absolutely super obviously AI. The opening paragraph is so riddled with AI-isms it reads like a parody of that. You have the raven like demon on a 'writing desk' that seemingly leads it to make a 'teapot' comparison because it thinks it's writing Alice in Wonderland. A claw tapping on wood is somehow like a composing a highfalutin 'symphony'.

#8 It's juggling a lot in a coherent way, but feels robotic about it. I'm surprised that this is not only human authored, but written by our famous Robin Hobb (who I've certainly heard of but never read).

>>24684134
Holy shit AI sycophant midwits are the worst people.
I DO use them. That's why I have the opinion I have. It lacks any creativity whatsoever. Which is why AI cultists like you will make a big deal out of how it can be better than some trite fantasy author but can never show an example of something writing by AI that is genuinely novel, interesting, or insightful. The nature of LLMs precludes them from being able to, as they are simply regurgitation machines. You can train it on all the fantasy novels you want, you will never get an output on the same level of quality as The Lord of the Rings. Not quite what you'd expect if it was making generational leaps in a couple of years as you laughably assert.
It's funny how there are still holdouts like you when the research is showing that AI isn't leading to mass job loss, or major increases in productivity, or really having anything near the impact you say that it is having, as most people are now toning down their rhetoric and "AI winter" is a term popping up more and more often.

>>24676176
That's not AI being good it's just Robin Hobb being a shit writer

>>24676987
>>24676274
Hobb is pretty bad. She's liked by r*ddit types who think her stronk female leads are interesting or fun when they usually aren't. Her books are quite dull and like all modern fantasy, way too long. Too many of them thinking making books longer than 600 pages means it's a good book.

>>24685256
>It lacks any creativity whatsoever
You're using hyperbole to say that it's not currently as good as humans in creative tasks, which I agree with.
But it's still better than a year ago, and much better than two years ago.
Why would its current stage be the way it will always be?
You have no reason to think that the current improvement trend won't continue to improve until it becomes good enough to match and maybe surpass humans.

>The nature of LLMs precludes them from being able to, as they are simply regurgitation machines
You have a very poor understanding of the technology AND of the way human creativity works.
Hint: humans are very advanced regurgitation machines too.

>You can train it on all the fantasy novels you want, you will never get an output on the same level of quality as The Lord of the Rings
Unlikely.
What makes great books great isn't some metaphysical magic force that humans can use to conjure up things. It's characteristics that can be analyzed and synthesized by software processes that are very similar to those that happen in the human brain.
Your baseless absolutist statements that AI can never do X or Y (despite it constantly changing and literally gaining abilities) is pure ideology.

>Not quite what you'd expect if it was making generational leaps in a couple of years as you laughably assert.
Again with the reality-denying. The generational leaps have already happened. Compare current ChatGPT with LLMs from 3 years ago. Or even better, 5 years ago. That's unquestionably a gigantic improvement no matter how hard you try pretending it's not.
You're the guy looking at the early Internet and thinking "this is too slow and impractical, it will never have an impact on the world" (on top of pretending it hasn't improved for some schizophrenic reason).

>It's funny how there are still holdouts like you when the research is showing that AI isn't leading to mass job loss, or major increases in productivity, or really having anything near the impact you say that it is having, as most people are now toning down their rhetoric and "AI winter" is a term popping up more and more often.
Yeah, if something doesn't happen immediately at a world-changing magnitude, then it can't ever happen.
Literally retarded.
A quick Google search shows that those things are indeed happening, only gradually because the tech is still in its early stages.
Imagine looking at the first cars and thinking that the automotive field as a whole can't surpass the horse just because the early cars can't.
>show me one car that can pull as much weight as my horse
>can your car run as fast as the fastest racehorse?
There's really no way for me to accurately articulate how comically stupid you are.