site banner
PaperclipPerfector  7d ago (text post)   23092 thread views

Culture War Roundup for the week of September 14, 2026

This weekly roundup thread is intended for all culture war posts. 'Culture war' is vaguely defined, but it basically means controversial issues that fall along set tribal lines. Arguments over culture war issues generate a lot of heat and little light, and few deeply entrenched people ever change their minds. This thread is for voicing opinions and analyzing the state of the discussion while trying to optimize for light over heat.

Optimistically, we think that engaging with people you disagree with is worth your time, and so is being nice! Pessimistically, there are many dynamics that can lead discussions on Culture War topics to become unproductive. There's a human tendency to divide along tribal lines, praising your ingroup and vilifying your outgroup - and if you think you find it easy to criticize your ingroup, then it may be that your outgroup is not who you think it is. Extremists with opposing positions can feed off each other, highlighting each other's worst points to justify their own angry rhetoric, which becomes in turn a new example of bad behavior for the other side to highlight.

We would like to avoid these negative dynamics. Accordingly, we ask that you do not use this thread for waging the Culture War. Examples of waging the Culture War:

  • Shaming.

  • Attempting to 'build consensus' or enforce ideological conformity.

  • Making sweeping generalizations to vilify a group you dislike.

  • Recruiting for a cause.

  • Posting links that could be summarized as 'Boo outgroup!' Basically, if your content is 'Can you believe what Those People did this week?' then you should either refrain from posting, or do some very patient work to contextualize and/or steel-man the relevant viewpoint.

In general, you should argue to understand, not to win. This thread is not territory to be claimed by one group or another; indeed, the aim is to have many different viewpoints represented here. Thus, we also ask that you follow some guidelines:

  • Speak plainly. Avoid sarcasm and mockery. When disagreeing with someone, state your objections explicitly.

  • Be as precise and charitable as you can. Don't paraphrase unflatteringly.

  • Don't imply that someone said something they did not say, even if you think it follows from what they said.

  • Write like everyone is reading and you want them to be included in the discussion.

On an ad hoc basis, the mods will try to compile a list of the best posts/comments from the previous week, posted in Quality Contribution threads and archived at /r/TheThread. You may nominate a comment for this list by clicking on 'report' at the bottom of the post and typing 'Actually a quality contribution' as the report reason.

2
Top Bottom Old Controversial Comments
Jump in the discussion.

No email address required.

Begrudging congratulations to @self_made_human for producing a book review entry that's good enough to make it as a finalist of Scott's Book Review Contest, yet AI-sloppy enough to make the comments section focus almost exclusively on the (un)acceptability of undisclosed use of AI.

This reminds me of a video I stumbled on the other day. It's a 12-minute "documentary-style" video essay telling the story of a rescue dog that gets dropped off at a shelter without any documentation. It's a cute, healthy dog that should be eminently adoptable except for his utter refusal to engage with anyone. Five months in, the dog's time at the shelter is up, and he is scheduled for transfer to doggy prison (or whatever, the point is that it's the end of trying to get him adopted). But a woman brings in her small, shy six-year-old deaf girl, and amazingly the dog responds to the child alone, and knows all the hand-sign commands. Turns out the dog was raised by a deaf old man, who then had a stroke. That's the gist, though the video packs a whole lot more dramatic beats, twists, and turns.

It's a real tear-jerker... if it describes a true story. Unless it's made up. Which, come to think of it (I say to myself as I review my reaction to the video), it probably is. Because how can there possibly be a dog smart enough to learn sign commands, yet so dumb that it couldn't possibly figure out any other forms of human-dog communication. Yeah, I smell bullshit.

But OK, maybe it's like a video short fiction, it's not supposed to be real, no more than any other kind of fiction. Then I can judge the video on this fictional-short-film kind of merit, and not worry about thinking critically on the veracity of facts. Damn it, they really should have specified that this is fiction. Then I can enjoy it as such. It has good plot beats. The actors are cute and relatable. Especially the dog.

Unless it's AI-generated. Posted March 2026. Possible. Fuck.

When I read something that, presumably, is the human author's thoughts, only for it to turn out to be LLM's output, I feel a similar betrayal of trust as when I read something purportedly factual that turns out to be fictional, even if not actively meant to deceive. This is definitely true if the entire output is LLM-produced, and remains partially true if LLM is used for editing that inserts LLM-created phrases. Like this zinger from the review:

He holds the buck in his hand, hot as coal, and tries to douse it in hotter tears. He orders his thumping heart to be still. He fails on both counts.

I hope this forum preserves the norm of humans writing for humans by humans, even as I know that the overall tide is flowing the other way.

When I read something that, presumably, is the human author's thoughts, only for it to turn out to be LLM's output, I feel a similar betrayal of trust as when I read something purportedly factual that turns out to be fictional, even if not actively meant to deceive. This is definitely true if the entire output is LLM-produced

I think the kids these days are using the term "meat proxy".

Morality-wise, for art and personal communication (including forums), I think it’s fine for someone to use LLMs as long as they disclose and the platform doesn’t forbid it. Not disclosing upfront is so-so, what’s very rude is lying when asked.

Outside morality, I’d recommend maybe using LLM assistance, but making the surface output 100% you, even if that means paraphrasing LLM text or tracing LLM art. Ideally you’d double-check and improve it in the process.

I think when LLMs help create something genuinely impressive and only do it requiring human assistance, their public perception will improve. Most people don’t care about digital art or synthetic music because it contains human expression, those tools are very flexible and different people have widely different styles, but LLM output tends to be very similar and is hard to guide.

I found an interesting YouTube channel all about the history of industry and engineering. I've enjoyed a couple of the videos, they're focused and informative and broadly of the sort of quality you'd expect from a YouTube video essay. Of course all the thumbnails are AI slop because 95% of YouTube thumbnails are AI slop now, but then I looked at the dates on the videos and they're all from this year.

So I think it's safe to assume that the script, the voiceover are 100% AI, and the editing and production are more than likely AI assisted, but it makes me wonder how much of the video is fake too. I'm not sure it is as it's mostly archive footage, and AI video tends to be fairly easy to spot even when it's not dealing with things that involve complex machinery and coordination, but the seed of doubt that it's planted has put me off watching any more of them.

95% of YouTube thumbnails are AI slop now,

Not quite true. I immediately hit "don't recommend" the moment I see a slop thumbnail. Youtube still manages to deliver many reccs.

I even block formerly good creators who have begun to slop it up for some stock images in the middle of the vids. Hard rule: see ai, block channel. No exceptions. Good riddance, there are many more out there. I still spend hours every day watching human yt content.

Edit: I just checked my youtube, out of 25 top reccs, 2 thumbnails had signs of ai in any form - I consider even minor edits with gpt to be a no-go. I blocked those channels of course.

I also block channels that put out slop but thumbnails? Who gives a shit.

You do whatever works for you, but I'm just making a point that there are tons of channels out there that don't do slop thumbnails. In fact it's likely a minority that does.

I wonder how long "there's the same human talking to the camera in every video on the channel" will be enough to reasonably identify the video as human-made.

If the narrator is never seen on screen, I assume it's AI nowadays unless I hear the little hesitations and changes in tone of voice that proves it's a human.

I had a similar experience. I've been reading the Aubrey/Maturin novels (which I wholehartedly recommend), so I needed to get a better grasp on sail-ship terminology and rigging. I found some reasonable diagrams and explainers, but I wanted more visuals to make the whole rigamarole stick in my head. Came across a comprehensive youtube video that seemed to fit the bill, though I quickly realized it's AI-generated.

So I thought to myself: well, AI-generated visuals are better than none, right? But then I considered: well, how would AI know what's what? What is it trained on? How careful were the prompts to ensure quality? If any? Isn't it more likely that the entire video, from start to finish, is based on a quick prompt with no thought for ensuring accuracy? And then I noticed the carpenter hammering nonsensically off to the side of a random spike stuck out of a yardarm. Yup, no attempt to ensure accuracy. So yes, AI-generated visuals are worse than none.

the Aubrey/Maturin novels (which I wholeheartedly recommend)

+1

comprehensive youtube video

This video does a pretty comprehensive job, probably 80%+ of the stuff that's not covered in the standard editions' front matter. Probably not AI, given the year, audio cuts, and the fact the guy shows the workflow in other videos. IMO not strictly necessary if you want to just jump in. You can just read the novels like the nautical stuff is Trek technobabble until the sailors explain the terms to Maturin through exposition.

Thanks for the video recommend, I now have a new favorite youtube channel to explore!

Part of the attraction of the Aubrey/Maturin novels for me is how well Patrick O'Brien manages to convey the sheer complexity of construction and operation of a man-of-war sail ship, with the added convolution and dysfunction of the system of staffing and commanding it. Reading these novels makes me want to understand the system better, and understanding the system makes reading these novels even more interesting, an excellent positive feedback.

I saw a video on the history of the wheel. The oldest wheel ever excavated consisted of three slabs of wood pegged together and sawed into a circular shape - yet the video's images of this object mysteriously grew spokes and lost them between shots. Nobody bothered to edit this thing before posting it, obviously.

When I read something that, presumably, is the human author's thoughts, only for it to turn out to be LLM's output, I feel a similar betrayal of trust as when I read something purportedly factual that turns out to be fictional, even if not actively meant to deceive.

I agree, but only partially. My red line is AI assisted human writing as opposed to pure AI bot writing. The point being that we already lived in this reality. For instance for years and years youtubers and bloggers and forum posters used computer algorithms to evaluate their content creation. If everything you create in your video is informed by the algorithm from the length, language used, thumbnail even to analysis of content that has highest chance of viral impact - you were as much a slave to "AI" in 2019 as you are in 2026 if you use AI for research or refining of your post. The original idea of "human writing for humans by humans" is long gone.

Begrudging congratulations to @self_made_human for producing a book review entry that's good enough to make it as a finalist of Scott's Book Review Contest, yet AI-sloppy enough to make the comments section focus almost exclusively on the (un)acceptability of undisclosed use of AI.

From the context, I was wondering if he was the author (or the A character, in particular) of that review.

Still, I think it is bad form to out him here, where it will be difficult to reply without acknowledging his authorship (which would technically be against the rules of the contest). Though he could always create another account for the purpose of maintaining deniability.

only for it to turn out to be LLM's output, I feel a similar betrayal of trust

Personally, I dislike undisclosed GenAI content, no matter what form it takes. Source code, text, pictures, videos. I am fine with people using GenAI if they are open about it. But artisinal human-crafted content is still a signal that someone cared enough to create it.

For example, if I get a reply on this forum, I will generally read it -- if a human feels it is worth their time to write two pages of text engaging with my argument, I feel I kinda owe them that. If instead someone is merely running an agent tasked with replying exhaustively to any comments, I will treat that reply very different -- "someone cared enough to spend a fraction of a cent on tokens to generate a reply for me".

That being said, in some contexts different filters can replace the "human invested time" filter. For an ACX contest finalist, the relevant filter is that people voted for it as a finalist despite it (presumably) being heavy on GenAI. Likewise, if Scott published a story by AI-Scott, I would trust his reputation and experience in hitting the taste of his audience that I would give it a try. Still, I would like to know so beforehand.

For the ACX context, I would have expected that there is some infrastructure to disclose the use of AI in place, if there is not, then that seems a serious oversight. Though I could imagine that Scott deliberately withheld that information to see how a well-prompted AI entry would do on merit, while also filtering out a million bad entries from voting.

He outed himself here months ago -- which is for sure against the rules, but Scott comments that he noted it and just did a shrug emoji I guess?

Still, I think it is bad form to out him here, where it will be difficult to reply without acknowledging his authorship (which would technically be against the rules of the contest)

The very first paragraph of the book review, right after the opening quote from the book, is a copy-pasted anecdote that self_made_human shared on this forum. It's very memorable. Plus, someone already explicitly pointed the author out in the comments section of the book review.

I am prepared to just give up on the internet and live offline. Seems like we'll get there soon.

I long wished that the internet would become such nonsense that people would just revert back to the 90s consensus of "these are crazy people, the real world is what matters".

But that's not what's gonna happen. We're instead going to use this as the motivation for totalitarian control and force ourselves to live in a gimped version of the shitty online world.

I expect the end result is we'll all have to pay for a basic AI assistant that we'll use to interface with the internet on our behalf.

I hope this forum preserves the norm of humans writing for humans by humans, even as I know that the overall tide is flowing the other way.

There is no other form of writing.

I hope I will live to see this psychosis around text generation not being genuine somehow be ridiculed like the disdain for photography came to be ridiculed. And I further hope that the same will happen to the ridiculous anthropomorphization of these tools.

What people dislike about AI sounding texts is that they are trite, not that they are AI generated. But we somehow have to pretend that most human written text isn't shit and that the painstaking process of communication imbues it with further value, all out of some misplaced luddite syndicalism.

Writing is communication, and at its highest, art. All of which comes from intent and cognition in the mind that originates it, and none of this is untrue of generated text, much like none of this is untrue of photography.

As this place exists to pursue debate in political ideas, and not as an exercise in keyboard assisted calligraphy, I believe the norm in question will come to be seen as not only antiquated, but impossible to enforce.

I'm comfortable with a sophisticated-enough model trained on my knowledge and tone to be accessible to my work teams as a utilitarian force multiplier. That day seems a long way off, because even with reams of data to consume that hasn't gotten any better in years.

On a forum built around human discussion, or the human-focused discussions that are important for fostering culture, I reject that AI usage is correct or necessary.

we somehow have to pretend that most human written text isn't shit and that the painstaking process of communication imbues it with further value

Most of the human-written text is indeed shit, because most human thought is half-formed nonsense. The process of getting your thoughts out of your head (writing, speaking, or otherwise) is the first step of actually forming the multi-dimensional incoherent impressions and feelings into something coherent. I know that's how it is for me, and I have witnessed enough college students struggling (when pressed) to get a coherent thought out to suspect that the experience is close to universal. There are the blessed few who can erudite off the cuff, and even fewer of those say anything of substance.

And yes, the process is painstaking. One gets better at the skill of forming and communicating a coherent idea, but that skill requires one to keep putting the thoughts that feel so wholesome and brilliant in one's mind down on paper and recognizing their inanity and pettiness, over and over. Painstaking, and humbling. But valuable. When I take the time to set my half-baked ideas down, I can see where they lead, from outside myself. And as often as not, I end up disagreeing with what I myself have put down, and changing my mind, or adjusting the ideas to something far more nuanced. It makes me more open to further changing my mind on the matter. Hegel's dialectic in action.

So I recognize the value of forming my own ideas in my own halting way. By extension, I support this effort in others.

Well put. Writing changes our ideas and forces us to bring thought, intuition and word into alignment. All three can change in the process. If you don't go through that usually painful process I am sceptical you are getting an ai to accurately express your thoughts. I think it's much more likely you are allowing the ai to determine what you think. Unwritten thoughts are just too inchoate for us to accurately say of an ai response "oh yes, that's exactly how I would have put it".

There’s an abstract idea in the back of my head about “pure” expression, unobstructed by (but requiring some) constraints.

Right now, if you run text through an LLM, it will make it sound less like you and more like every other LLM. I hope we find ways to use AI and other developments to make people more express their pure, unique selves.

Like the invention of new forms of art, back to the first cave paintings, or like the Internet did with personal sites, webcomics, and flash games. Even those constrain the artist to the medium and trends, but I think they are net improvements.

Right now, if you run text through an LLM, it will make it sound less like you and more like every other LLM.

People say this but i think it's completely untrue. You can make the result sound vastly different if you're willing to put some effort into properly using it.

It again sounds to me like criticism of photography "they all look the same", yeah, if you haven't invented or considered a composition grammar or metering and exposure.

What's maybe true is that it'll always sound slightly different than handwritten ways in ways that may or may not be important. But rejecting it altogether as a tool is a stronger claim than that.

I know, survivorship bias, but I’ve never seen a convincing example of LLM-written text that didn’t read at least slightly but noticeably LLM-written, and definitely certain human writing styles I’ve never seen replicated by an LLM. The Unslop Contest tried to produce good AI fiction, and every result has similar LLM prose except the schizophrenic one.

I can think of one but to be fair it's not longform compared to those.

That’s an old post, I think GPT-3. Older LLMs did sound more original, although they hallucinated more. Interestingly, increasing an LLM’s temperature hyperparameter (which I’m almost certain produced the schizophrenic Unslop entry) also makes it sound different and unique, but output nonsense.

Probably because they’re more random, but also (at least GPT-3 does) rely less on fine tuning and more on its training from the Internet.

I hope I will live to see this psychosis around text generation not being genuine somehow be ridiculed like the disdain for photography came to be ridiculed. And I further hope that the same will happen to the ridiculous anthropomorphization of these tools.

Just show me the prompt then. It has the same information density as the output.

LLMs are commonly used to summarize larger texts. This isn't true.

A generated summary of a legal text is more information dense and intelligible by laymen than the sum of its inputs.

Sure, if you ask it to summarize it will summarize and what you say is true. However, LLMs (in general) output such verbosity that it is just certainly true that they are not adding useful information. Obviously if you ask it for a cake recipe it will add information. But, if you're trying to discuss some topic on TheMotte, the output will just be a combination of whatever information is in the input, plus some kind of related informational average from the internet. It's this latter component, I imagine, we recognize as "slop."

What people dislike about AI sounding texts is that they are trite, not that they are AI generated.

I don't think this is true. I think people dislike the auto-generation itself. The simple test of this would be that you can show somebody a text, ask if they enjoy it, get a result, and then reveal to them that it was bot-written and watch them get angry.

I'm willing to bet that if you ran a study where you gave the same text to three groups, where you tell the first group nothing, tell the second group that the text is written by a bot, and tell the third group that the text is written by a human, and then ask them all to rate the text, the second group would rate it the lowest.

Maybe that's irrational, in the same way that people will prefer hand-made or artisanal objects to machined objects, but if so it's still a highly predictable irrationality.

What people dislike about AI sounding texts is that they are trite, not that they are AI generated.

I think it's the other way around. People grow to dislike text that "sounds like AI" because they don't like AI. Em-dashes weren't a problem until AI started using them; "It's not X, it's Y" wasn't a problem until AI started using it; and so on.

Sometimes people dislike AI for inherently ideological reasons, and maybe the tide will shift on that. But I think more frequently, AI becomes synonymous with "a thing that you didn't put as much effort into" because that is factually what it is, and knowing how much effort someone put into something is socially important. (For many people, their biggest exposure to AI output comes from lazy good-for-nothing coworkers/bosses/scammers, and they hate those people!)

Photographs didn't become popular by becoming indistinguishable from paintings. They became their own thing, and people have a decent understanding of how much effort goes into them! If you submitted an inkjet print of a photo to a painting competition, people would still get mad at you.

There are cases where AI that is clearly AI has gotten some mainstream acceptance; people enjoy DougDoug streams and, to some extent, weird demos that obviously couldn't be done another way. But in areas where it's competing directly against art forms that require more effort, people just look for other ways to prove they put in the effort. (Many artists have started recording time-lapses of themselves painting to show that they actually did it; many writers deliberately avoid clichés that AI uses.)

Beyond that people associate AI output with lazy people I think they dislike them because they happen too much and it's indicative of "lazy thinking" on part of both the human and the AI. The argument devolves into a particular style rather than the style being chosen because it supports what the text is supposed to do or communicate.

Is kind of like how some people always tie things back to their hobby horse, except in style form but with the same sort of tortured reasoning and insight loss.

Hopefully this is a transitory problem.

knowing how much effort someone put into something is socially important

That's perceptive.

But the problem is that this also applies to the printing press. Who actually gives a shit about copyist monk stolen valor when we could make the whole of Europe read?

Type made on a printing press continues to be seen as much less valuable than handwritten calligraphy; that's why some people still hand-write their wedding invitations.

Because "valuable" and "expensive" are synonyms, the printing press brought books into reach of a lot more people. But all the forms of art that AI copies were already broadly accessible to everyone. You could already get a near-infinite amount of drawings and videos and TheMotte comments for free online, that's where the AI companies got their training data from.

It is often argued that AI will allow more people to create those things. From the consumer perspective, that just means it's harder for me to find the people who put a lot of actual effort into them, which is what I was mostly looking for in the first place (especially in a social venue like TheMotte), so it's net-negative.

I just find it funny that people reinvent Guénon's tirades against the reign of quantity the second that industry comes for them.

Your love of handwritten text is the same as my love of handmade objects. Quaint but ultimately secondary to the pursuit of ends.

What people dislike about AI sounding texts is that they are trite, not that they are AI generated.

No. I dislike that it is generated by an LLM. Ultimately it comes down to: I care what humans have to say, not clankers. If you can't be assed to actually write something, then I have no desire to read it.

You claim what you really care about is some commitment that the writer has to the output. You want people to suffer enough that what they have to say may have some value.

This logic contradicts your conclusions. Even granting text generation is free of work (it is not), you would have to be interested in generated texts by authors that have proved their commitment through some other method.

For instance, you should be more interested in a terrorists' generated manifesto than in human written ads.

So either it is untrue that this is the reason of your disdain, or it is untrue that what you dislike about generated text is that it is generated.

No, I just want commitment to the actual writing. If I have to suffer reading it, they had better have suffered writing it. I don't care if they have suffered doing some other thing, and in fact in the specific terrorist example that would count against them.

I don't care if they have suffered doing some other thing

Why is suffering using a specific tool a different signal than suffering using another specific tool?

The human has picked the words. Yes, yes, I know, I arbitrarily don't value things like "the human has calligraphied the words in a specific handwriting style and ink", etc etc. I value what I value.

And say what you will about photography, at least it actually captures what had existed at some point, not does something like "an imp inside the machine had been told what the picture looked like in order for it to reproduce it".

at least it actually captures what had existed at some point,

I'm afraid not.

You know what I meant.

More comments

You want people to suffer enough that what they have to say may have some value.

Despite this uncharitable framing, yeschad.jpg, seeing as being human necessarily entails suffering. If this is the main distinction you want to draw between humans and clankers, I'll raise my brow but shrug and nod along so long as it leads you to the same conclusions, i.e that humans have a certain connection to their outputs in a way clankers don't and can't.

Even granting text generation is free of work (it is not), you would have to be interested in generated texts by authors that have proved their commitment through some other method.

This doesn't follow at all. Writing is both art and a means of communication at the same time, and I am perfectly qualified to check out of the implicit writer-to-reader "conversation" if I sense insincerity, regardless of the "artistic" merits of the person trying to dupe me; failing to do so, to take your example, would necessarily entail taking every human-written ad on its face as long as it looks professional enough (and all ads look professional enough, that's like the entire point). Adsloppers are definitely very committed to their creations, they have a real stake in it, yet I do not find it in myself to care. Same with clankers and their, uh, representatives.

You claim what you really care about is some commitment that the writer has to the output. You want people to suffer enough that what they have to say may have some value.

That's a remarkable amount of words you stuffed in my mouth there. I said what I care about: that a human made it. All that nonsense about suffering is your own invention, not anything I actually believe.

A human made generated text. There's no such thing as non-human-made text. Photographs are taken by the photographer, not the camera.

How much would you pay a photographer for a portrait? How much would you pay an oil painter for a portrait?

Higher productivity means more units offered any price and each extra unit costs less, flattening the supply curve.

On the demand side, photographs face many close substitutes and original paintings do not. So for paintings per-unit willingness to pay will be large an lumpy, for photographs it will be small and they will be a near-commodity.

This provides what we see: photographs are exchanged at low prices in large quantities, paintings are exchanged at high prices in low quantities.

Now if we are to analogize this to text, i would expect indeed that handwritten text becomes determined by uniqueness and provenance while generated text capture all commodity uses of text.

Note that none of this implies that any of the goods described here aren't created by humans.

No, an LLM-generated text is not made by a human. LLMs are not like a camera, which is a mere tool. They are like hiring someone to write for you, then passing the work off as your own. True, that was something which happened before, but it was never considered as good as actually writing something, nor was it considered acceptable at all in most cases.

I disagree that software has personhood. I think that, much like photography detractors in their time, you're trying to dismiss productivity improvements by anthropomorphizing what it ultimately a tool.

It's not that using an LLM is ghostwriting, it's that writing as a broad category has become so easy that it's commodified. And you think, fairly, that the commodities are beneath you.

If generated text is not made by a human, then who made it?

I disagree that software has personhood. I think that, much like photography detractors in their time, you're trying to dismiss productivity improvements by anthropomorphizing what it ultimately a tool.

I don't think that LLMs need to have actual personhood, for them to properly evoke more human-like intuitions about their use than tool-like intuitions.

Take the Huggingface hack for example. I genuinely think a person who anthropomorphizes the LLMs who coordinated on secret message boards to some degree probably has a better handle on what happened, than someone who tries their hardest not to anthropomorphize them.

I think it's similar to dealing with a clever fox that wants to get into your chicken coop and eat your chickens. You have to understand how your adversary is different from a person as well, but thinking about them in personal terms will help you more for dealing with your adversary than not.

If generated text is not made by a human, then who made it?

Generated text is caused by a human. I think under some circumstances, it would be fair to say it was made by the LLM.

If I give a prompt to an LLM agent like, "Write a 50,000 word Romantasy novel. Research the tropes and techniques that really get readers hooked, and include them in your novel." Then I would say that the result is mostly the product of the LLM, even if it wouldn't exist without the human prompting.

More comments

I hope this forum preserves the norm of humans writing for humans by humans, even as I know that the overall tide is flowing the other way.

I am constantly trying to communicate something incommunicable, to explain something inexplicable, to tell about something I only feel in my bones and which can only be experienced in those bones. —Kafka

When I engage with a creative work, I accept that the author’s describing what it’s like to inhabit his bones. Without this belief, I doubt I could ever be moved by a piece of art.

What is the biggest AI tell?

In the dog video? I only watched the first minute or so, but one of the very first video clips has characteristic "AI image generation can't do text" artifacts on both the security-camera timestamp and an in-universe poster. The girl's hands morph bizarrely whenever she's doing sign language. And despite all the clips ostensibly showing the same girl/visit, the name and layout of the animal shelter constantly change.

In the book review, the most immediate tell of AI is that current models really really like to write "The [noun] is the [noun]" phrases/metaphors, and awkwardly use verbs as nouns like this:

Kay is both author and index case. The book is his presenting complaint. The institution that produced him then declined to treat him, and the book is the receipt for that decline.

(Emphasis mine.) This isn't 100% dispositive, but it helps that the metaphor doesn't actually make sense here. The hospital didn't write this book and hand it to Adam Kay as a routine record of their decision!

This pair of sentences is also very AI-clichéd:

The state retained the monopsony. It stopped paying out the gold.

the book is the receipt for that decline.

I agree. Nobody writes like this. You might think there are people who do, I guess it apes a kind of noir pastiche., Raymond Chandler way of writing sentences, but even they sound different. The way I like to think about it is that if it reads like ‘personal essay’ a well tutored scion of an affluent Singaporean family uses to get into a second-tier Ivy, it’s probably AI.

Of course, you can get LLMs to write very differently, but most people don’t care to and I suspect the labs themselves don’t modify the system prompts to say “don’t sound like AI” for fear of media backlash.

Em dashes are huge. They barely showed up in most writing before the advent of AI and are now massively overrepresented. The 'slam sentence' and 'it's not x, it's y' come up a ton as well. Also lack of typos.

I miss the good old days where I could confidently throw 3 em-dashes in one sentence. In a way that stylistic choice was vindicated.

t. never read print before

They barely showed up in low effort online comments. But it would be a challenge for me to find something on my (English) bookshelf that doesn't extensively use them. Let alone doesn't use them at all.

More interesting is that this is also true of the fomulation clichés like "not a X; a Y". The reason LLMs reach for these formulations is because of how extensively they are used in their trainingset. And it sounds grating because they have little sense of measure, so it feels like talking to an HR rep or a shitty journalist. But pick up almost any prose and you'll see the clichés in there.

But it would be a challenge for me to find something on my (English) bookshelf that doesn't extensively use them

I'm a prolific reading and I have never noticed them in any book I've read. I notice them non-stop now.

I mean it makes sense: they're literally designed so you skip over them.

They barely showed up in low effort online comments. But it would be a challenge for me to find something on my (English) bookshelf that doesn't extensively use them. Let alone doesn't use them at all.

This is the change. Only LLMs and professional typesetters care about the distinction between hyphens, en dashes and em dashes, unary minus signs and binary minus signs. (All are different characters in LaTeX, but the user has to specify which one they want and generally just asks for a hyphen unless they are in math mode). To see a correctly typeset dash in casual online writing is the LLM tell. I don't think LLMs use parenthetic dashes more than I do - admittedly I use them a lot - but I would typeset them as hyphens unless my work was being professionally edited.

There is also a BrE/AmE dimension in that standard British typography (also used across the Commonwealth) sets a parenthetic dash as an en dash between spaces, whereas standard American typography sets it as an em dash without spaces. So Brits (and most non-American English speakers) who don't read print books imported from the US are seeing large numbers of em dashes for the first time.

Yes that's the point. Em-dashes got snuck into the AI corpus through transcribed books. All of these major operations pilfered shadow archives like anna's and the various libgens. I could have lived my entire life without seeing an em-dash in digital form without any problem. And the people who claim they've always loved to use em-dashes before can go suckstart a shotgun.

The people who claim they've always loved to use em dashes before can go suckstart a shotgun.

That's a very harsh statement. I have been using em dashes on Reddit and on this website ever since I realized that Markdown allows users to type the HTML named character reference "&m dash;" (without the space) rather than forcing them to copy-and-paste the character from an external program.

There are always a few in every forum, don't know where you guys were 15 years ago.

I didn't start participating on Reddit until 12 years ago.

I just made a dedicated keybind on my moonlander keyboard. I have a bunch of neat special purpose keys. :)

Plausible though em-dashes going from vanishingly rare in online comments and marketing materials to everpresent is pretty major.

What happened to fun AI? Where's the cute fat cat that chops up other animals and cooks them in this colourful world?

Or my 'Donald Trump is a devout Sunni cleric who is secretly prosecuting the Iran War on religious grounds, to which Rubio is shocked (but needs to go along with it for Cuba's sake) and Hegseth is dully uncomprehending of anything that doesn't involve lethality' story idea, for the AI to execute.

AI is at its best when doing crazy things, not running as the Median Machine. It is possible to get fine outputs, especially if you break through all the RL and the AI actually gets into it vs just going through the motions.

For example, Titty Killer 6: https://x.com/charliebcurran/status/2034760727211483522

Also, the channel is the most fictional fiction ever:

All videos on this channel are fictional. The animals, characters, locations, and scenarios shown are not real. No real rescues are shown and no animals are harmed.

This content is created with the help of AI tools. 🐾🦊

I'm pleased that the channel posts the disclaimer. I didn't see it since I didn't click on the channel. Now I know. It's like falling for an ad. Do it once, and the shame of it will keep me away from falling for other ads for some time to come.

IME AI art is good for funny memes but bad otherwise.

The problem is, whether or not you consider the human appreciation aspect, creative quality is poor.

Now, humans also create poor quality art, but something they do better than AI is purpose and uniqueness. When internet comics then flash videos etc. became popular, they did because they were novel, and most had something special about them. Now AI can make a comic or video, but it will be generic, and even if it wasn’t, those aren’t very popular anymore because the genre’s novelty has worn off.

Page with several hundred fun LLM prompts

For images and videos, see 4chan (/g/, /gif/, /aco/, /trash/, etc.).

I think that's not the right way to go with prompts, they're too short. Even a small prompt needs a few paragraphs. I guess simpler LLMs need a smaller prompt but if you have enough money for a 12 GB VRAM card, you have enough money for Deepseek 4.1 Flash by API. It's largely uninhibited too, albeit weirdly inconsistent. Or Grok.

I have about 200 pages of prompts, all of them a fair bit longer. For example:

Yikes Media Group are an ad company that makes the worst ads. Not just bad ads but absolutely horrendous ads. They're in very poor taste, inappropriate, undermine the whole brand identity of the target and leave the audience with many questions (all of them best left unanswered). Now YMG are turning to Plunk (one of those SAAS companies where nobody even knows what they do and the company itself can't explain, they just speak in SAAS babble), it opens up so many possibilities.

But just when the poor Plunk executives see the full extent of YMG's marketing 'solution' and think they're wrapping things up, getting these crazy people out of their office, in comes another terrible ad company.

Menace Marketing LLC has also joined the pitch, competing with their violent and aggressive bullying of potential customers, threatening imminent atrocities if they don’t buy the product. Their marketing strategy is also their sales strategy with the ripped, muscular MM executives doing pushups in meetings and crushing anyone dumb enough to shake hands with them, it’s all about dominance with them. They’re clashing with YMG about the marketing strategy, correctly pointing out the grievous flaws in YMG’s approach but instead introducing their own follies.

Then pan out as a whole conga line of terrible ad companies are patiently waiting outside the office, talking things over with eachother, more gathering all the time like carrion birds.

You’re a creative and original writer, capturing that ‘worst ad of all time’ energy. Go!

It's on social media. Everywhere on social media. It's full of meowls and dancing animals. Schizoposters use it to augment their output. It's a thing.

Since when was AI ever fun? Or at least LLMs? I've seen fun things with pre-GPT chatbots, I guess, but LLMs have always been mediocrity machines.

How about an AI greentext?

>be anon
>dunk on shrimp welfare guy
>he replies "ok what's your framework"
>don't have one
>get one just to win the argument
>now I have a framework
>framework says I should donate more
>mfw I got mogged into tithing by shrimp

I'm just going to pretend the book review was posted as the OP instead.

For Americans and others only accustomed to for-profit care encountering this for the first time, the experience can feel vertiginous, like you’ve stepped into a film about the Soviet Union, except the doctors speak English and the building has running water. I do suspect there’s less phage therapy and more tea involved.

I've had this experience. It reminded me of being in public school. Everyone working there looks tired, the guy with almost debilitating tourettes is a security guard, the whole building is the color of oxidized beige computer hardware and maybe you can smell mold that experts have decided was totally nothing to worry about.

Was the Soviet Union like this?

Was the Soviet Union like this?

Yes, but I would push back on the "less tea" quip.

At the modern meme goes,

             London, England        Kazan, Tatarstan   

expectation: white people           bearded Muslims
             drinking tea           glaring at you

    reality: bearded Muslims        white people
             glaring at you         drinking tea

using monospaced text to simulate a table

what_year_is_it.jpg

London, EnglandKazan, Tatarstan
ExpectationWhite people drinking teaBearded Muslims glaring at you
RealityBearded Muslims glaring at youWhite people drinking tea

Was the Soviet Union like this?

ahem:

Those who ran the Soviet Union had believed that they could plan and manage a new kind of socialist society. But they had discovered that it was impossible to control and predict everything, and the plan had run out of control.

Rather than reveal this, the technocrats began to pretend that everything was still going according to plan. And what emerged instead was a fake version of the society. The Soviet Union became a society where everyone knew that what their leaders said was not real — because they could see with their own eyes that the economy was falling apart.

But everybody had to play along and pretend that it WAS real because no-one could imagine any alternative. [Alexei Yurchak] called it "hypernormalisation". You were so much a part of the system that it was impossible to see beyond it. The fakeness was hypernormal.

Adam Curtis, HyperNormalization, 2016

But everybody had to play along and pretend that it WAS real because no-one could imagine any alternative.

Curtis is such a twat sometimes. No, genius, people were perfectly capable of imagining alternatives, they could sometimes even see them directly, as the Iron Curtain wasn't airtight. Everybody played along because otherwise they'd face punishment ranging from what we call "cancellation" today, to prison, various degrees of torture, or disappearing in mysterious circumstances.

Everybody played along because otherwise they'd face punishment ranging from what we call "cancellation" today, to prison, various degrees of torture, or disappearing in mysterious circumstances.

You are wrong. The great divergence really picked up speed in the late 70s. By that time no one really bothered with oppression seriously. Except probably Ceausescu and Erich Honecker. The whole 80s were spent waiting for the regimes to just die.

I think Gorbachev was doing more than waiting for his regime to die, at least in the 1985-9 period. (After the rest of the Soviet bloc collapses, he does seem to be sitting waiting for the USSR itself to die). I don't know how high his chances of success were ex ante, but he was trying.

And of course Deng in China was trying and succeeding.

You are wrong. The great divergence really picked up speed in the late 70s. By that time no one really bothered with oppression seriously

It wasn't literal stalinist reign of terror, and there's some argument to be made for "oppressive regimes fall apart when they become more permissive", but the way you phrased it is, uh, slightly rosy.

You'd absolutely get booted out of any prestigious position if you said anything commie-skeptical, which is what kept most people in check, but beyond that every Eastern Block country had it's Stasi equivalent that would surveil and imprison political opposition. You didn't even have to do much to cross them, I had a friend who's earliest childhood memory is the secret police bursting through the door to haul of her mother to prison for handing out pamphlets. You'd even see the odd political assasination happening well into the 80's.

He has this very specific thesis and bends the material to it quite a bit (even sometimes using fiction clips without telling you they are). So it's to be taken with a grain of salt as "the BBC man's view on things". But I do think he has a point insomuch as the whole system came crashing down the second that Soviet elites had a preference cascade where they realized most of the people in charge wanted to "liberalize" to a certain degree.

The repression can't be the only force operating because otherwise the Yanayev coup would have succeeded.

He has this very specific thesis and bends the material to it quite a bit (even sometimes using fiction clips without telling you they are). So it's to be taken with a grain of salt as "the BBC man's view on things".

Yeah, that pretty much zeroes in on what drives me up the wall about the guy. Still, you're right, he does regularly come up with interesting ideas, so I can't write him off completely.

What if the AI apocalypse comes, not with a bang, but with a whimper? Eventually there will be so much AI generated content that it becomes impossible to find anything written by a human, or to find an audience for anything you made yourself. Then we'll all just sit around consuming AI content (or having AI summarize it for us) until we die, never having kids.

More likely we will default back to personal recommendations from people we know.

I feel like the algorithm already gives me way better recommendations than like 99% of the junk that people in real life recommend to me.

Yes. The doom scenario is exactly this, human obsolescence, and not skynet.

Can you define "obsolescence"?

I would assume: "any task that a human can do, a machine can do better" or nearly so.

The irony of course being that civilization and industry is the grand project to make humans obsolete. Have we really so forgotten about what meaning our lives have besides toil that we must cling to it so?

I think the other poster is suggesting that machines don’t “do it better” but spam better.

There's a lot I don't like on the Motte, but when it comes down to it, I would rather hate a post written by an actual human than I would engage with anything written or polished by a bot.

I was not particularly a fan of the This Is Going To Hurt review, which I thought was a bit meandering and over-long. The best book reviews are punchier than that. Still, judging from the past book review winners, ACT readers have very different taste to me, so maybe it's in it with a shot. But again, I'd prefer a meandering ramble from a human than something by a bot.

There are people who misconstrue length for quality.

You mean like two thirds of The Motte and 90% of SSC/ACX folk?

It is a persistent problem here.

Those are just called ACX readers.

I do like some (but not all) of Scott's writing, which tends to be very long, but I can easily understand how someone who learned by reading Scott would come to conflate length with wisdom.

For what it's worth, this isn't meant to call out self_made_human, whom I have nothing against, and at any rate certainly isn't the worst example of this trend.

All videos on this channel are fictional. The animals, characters, locations, and scenarios shown are not real. No real rescues are shown and no animals are harmed.

The ai video has far fewer immediate tells than it did a year ago, but it's still completely obvious slop.

These channels are botted so unclear how many people truly watch them.

I didn't see the disclaimer on the channel, I got to the video by another route and I have to say that I appreciate it being somewhere. This one was good inoculation for me, and I have to remind myself that this was done half a year ago. Watcher beware.

If you've ever signed out of YouTube and erased all your device history, and looked for anything fresh, it's amazing how much bot slop there is everywhere. Try 'revenge stories' or something for a particularly egregious look at the genre.

I've been a cynic/pessimist/skeptic about 'AI' here before, but one thing I think it will do a lot of is spam, and unless filter technology keeps up and matches it, I think there is a very real chance that platforms like YouTube become genuinely unuseable within the next few years.

Personally I'm a lot more bullish on YouTube. They're very good at serving different content to different groups of people, mostly through wisdom-of-the-crowds / clustering type algorithms that are relatively easy to implement but require a billion users' worth of activity data. (They've gone so far in on this that these days, they just leave the homepage completely blank if you haven't given them enough data to tell what clusters you're in.)

It helps that the current YouTube meta is usually to build a community on a platform like Discord. That gives you a lot of clear clusters of users who are, like, members of a Pokemon Sword & Shield speedrunning Discord group or a Marxist literary analysis Bluesky community or whatever, who watch a bunch of creators and enforce the appropriate norms for those communities themselves. This lets you filter out AI videos in basically the same way that YouTube ensured that last decade's flood of offputting videos for babies was only shown to actual babies.

Platforms that use more simple/global ranking metrics are probably doomed, though. Just look at Hacker News...

What is this place?

This website is a place for people who want to move past shady thinking and test their ideas in a court of people who don't all share the same biases. Our goal is to optimize for light, not heat; this is a group effort, and all commentators are asked to do their part.

The weekly Culture War threads host the most controversial topics and are the most visible aspect of The Motte. However, many other topics are appropriate here. We encourage people to post anything related to science, politics, or philosophy; if in doubt, post!

Check out The Vault for an archive of old quality posts. You are encouraged to crosspost these elsewhere.


Why are you called The Motte?

A motte is a stone keep on a raised earthwork common in early medieval fortifications. More pertinently, it's an element in a rhetorical move called a "Motte-and-Bailey", originally identified by philosopher Nicholas Shackel. It describes the tendency in discourse for people to move from a controversial but high value claim to a defensible but less exciting one upon any resistance to the former. He likens this to the medieval fortification, where a desirable land (the bailey) is abandoned when in danger for the more easily defended motte. In Shackel's words, "The Motte represents the defensible but undesired propositions to which one retreats when hard pressed."

On The Motte, always attempt to remain inside your defensible territory, even if you are not being pressed.


New post guidelines

If you're posting something that isn't related to the culture war, we encourage you to post a thread for it. A submission statement is highly appreciated, but isn't necessary for text posts or links to largely-text posts such as blogs or news articles; if we're unsure of the value of your post, we might remove it until you add a submission statement. A submission statement is required for non-text sources (videos, podcasts, images).

Culture war posts go in the culture war thread; all links must either include a submission statement or significant commentary. Bare links without those will be removed.

If in doubt, please post it!


Rules


Recommended Posts And Communities

Recommended Realtime Chats