AI Fiction Is Easy to Detect Because It's Stupid and Bad, Research Finds

I learned recently that it was actually Twilight fanfiction to start, and the author had to tweak the world to make the vampires rich guys.

Which I guess explains it a bit more.

Which is pretty telling of how bad llm fiction is.

50 shades of grey and fanfics like my immortal are fun to read because a person was weird enough to write them. Llm slop is just spam.

50 shades of gray was very gray tho

that is a very good explanation of why LLMs generate words which contains no meaning

Yeah, AI sucks for writing, but I had a back and forth going where ChatGPT wpukd produce like, 3-4 paragraphs and dialog and I would tell it where to go next and it was producing some pretty good stuff. This was back on like, 2 or 3 model versions ago too. I had to stop because it updated the model and suddenly it was all "Wait, you had me writing a sexy story!!!! I can't do that, I am programmed to be a puritan!!!!"

You are not going to get a whole book ornshort stoty with one prompt.

Not explore, not search, but delve!

Excuse me while I wash my hands after typing that.

How am I supposed to write a story about goblin miners now?

You are wrong, even a LLM fed with 26 great books couldn't write a good one.

Rule of three is a other one. Although, it does make the output nicely structured.

Load bearing

there's stuff and it happens and the characters are just kinda there

You're welcome to ignore this, but I feel I have to point out, not to you, but the universe in general (and just coincidentally as a reply to your comment), how well this describes almost everything about the most recent two episodes of a beloved TV series featuring a blue box that travels in time and space (which I like the usual amount). I'm going to go cry now. 🟦

Ugh you're right about Duolingo stories. They're so formulaic nowadays. Half of them are about Oscar brown-nosing some journalist.

Now im wondering what brand of slop its put out for the illustrations

You're absolutely right!

/s

LOL I mean human slop has always been a thing, so this is unsurprising. I at least find it to be highly memeable and fun to mock casually. However, slop engines outputting garbage a human never touched, that is a different matter.

The LLMs learn to write fiction from the best[-seller]! 🤣

I mean at least reading slop like 50 shades of grey or shadow of the conqueror can be amusing as a gateway at someone's psyche. I love bad fiction readings when it's human; when it's and Llm there is this odd creeping feeling that it's a waste of time. Same for bad music, bad cinema or bad art. If an llm made something like the room or fateful findings, it would not be amusing at all.

Given that people seem happy to consume endless A.I. Instagram feeds and such, I get the feeling most people actually don't care about the quality of what they're consuming. People who hate A.I. content like myself seem to be in the minority. I've actually run some casual anecdotal triangle tests with friends and family, and almost everyone actually preferred the A.I. generated content.

To me it seems almost like the equivalent of fast food or hyper palatable food maybe. It's shiny and bright and obvious and blunt and patronising and sycophantic… but people seem to enjoy that. I'm not sure how I feel about this.

Which is why they looked at 50,000 stories for broad trends, not just one

They let an AI do the evaluation.

How dare you make me use a search engine? /j

Sturgeon's law states, "Ninety percent of everything is crap". The adage was coined by American science fiction author and critic Theodore Sturgeon while defending the merits of the genre. Sturgeon observed that most works in any field were low quality. Therefore, science fiction was not uniquely inferior.

Source: Take a wild guess

Since LLMs try to find the most lilely word combinations, I'd say it is actually impossible for an LLM like we have them now, to write a surprising or exiting book.

There are compelling reasons to think that even in principle, these digital models can never produce good literature.

https://www.psypost.org/a-mathematical-ceiling-limits-generative-ai-to-amateur-level-creativity/

When I write fiction, I’m attempting to evoke vicarious impressions or to evince something strange about the world, where my success depends on the novelty and efficacy of my metaphors (for instance) to convey subjective experience.

This goal is grounded in reality. We are embodied creatures after all.

LLMs are not grounded in reality. They are trapped in language. They are not capable of genuine novelty because they can neither go outside nor generate interesting experiences worth communicating.

This is why they suck and will probably always suck when it comes to art. It’s also why LLMs are super good at writing scientific, technical, and business documents, all of which specifically eschew subjectivity in favor of formulaic chunks of text.

The more we criticise muppets, the more they will take notes and improve. And before we know it, we won't be able to tell if a movie is human acted or muppet acted.

"It will take notes"? AI is just autocorrect on steroids.

the wilful ignorance and denial is really just the unspoken feeling of uncanny valley surfacing

This article is discussing a research paper that looked into AI's hilarious tendency to "delve," among other things, and one of the many conclusions they came to is exactly what we all already suspected: it sucks at fiction. That is the furthest thing from being willfully ignorant.

Yes, we know that LLMs can "imitate art", but regardless of how good an LLM gets, it's still going to be an LLM. LLMs aren't AI, no matter how much they improve, because for that you have to have understanding, not just a statistical model.

I do have good news for you, though. There are several open-source LLMs.

They can't even reliably get image generators to stop generating people with six fingers.