I strongly disagree, for reasons that are different from what's discussed in the thread.
It doesn't matter if LLMs can't hold a candle to some of the finest human writing. What matters is that before you become a world-renowned writer, you need to pay your bills, often for a long time. And that's what LLMs take away. All the mundane literature-adjacent works: journalism, translations, technical writing / corporate comms, copyediting.
The same goes for many other forms of art. A budding painter or a musician could support themselves off commissioned / commercial work while working on their grand opus... but now, the customers just prompt gen AI.
The first careers I witnessed being negatively impacted by AI coming onto the scene .. just a couple years ago .. was tech writers. Source: my personal network of tech writer colleagues.
Have seen the same thing happen to literary translator friends. Barring the very rare few who broke through to international fame via well-known literary award, the rest were largely sustaining themselves with more mundane business translations and that work is ended.
My father knows a guy that financed a Texas mansion, for $2MM, using the collateral from his decade of running a very large technical document translation service (US resident, using foreign workforce from his homeland), popular throughout the 2010s.
Was a beautiful house, which I briefly lived in during scatterbrained renovations (never met the new owner).
The bank forclosed on his house ~late2024/early2025... anybody can now just use an LLM to translate (in real time), on sub-$300 hardware (offline, open-source models run via e.g: Whisper[GUI]).
Yeah not seeing it for technical writing. Especially in industries with even a hint of regulatory compliance, the imprecision is too significant. It's not for lack of trying or wishful thinking.
Either your colleagues are facing stiffer competition off-shore (not LLM driven) or their jobs were more about organizing, checking and sorting topics and data (LLMs can take that away)
AI will definitely not take over translation for a long time. Machine translation has improved to the point it's mostly no longer incoherent, but for non-formulaic content it's still very hit-and-miss, not that much better compared to 10-20 years ago. For example, something they still struggle with is consistently translating terms coined within the text with consistent phrases.
>A budding painter or a musician could support themselves off commissioned / commercial work while working on their grand opus... but now, the customers just prompt gen AI.
Are you joking? The only people I see using generative AI are either companies, shitposters, and spammers. Who are all these people who have stopped paying commissions and moved over to prompting AIs?
There's another translation use case worth considering. For non-native English speakers, AI can help express ideas they already have in a language they're less comfortable with.
Especially on forums like HN, the AI isn't generating the person's views - it's helping them say that what they already mean more clearly in English.
But what share of translation dollars are non-formulaic content? To me it seems that there are tons of manuals for dishwasher and such to translate (where from my experience quality basically doesn't matter) and only a limited amount of diplomatic communiques.
Manuals have been machine-translated for ages, well before algorithms were anywhere close to ready for the task. No change there.
>a limited amount of diplomatic communiques
With zero data to support it, I'd bet good money the vast majority of semi-professional translators (that is, those not employed by publishers but still making some money off of their work) work on fiction, translating comics, subtitles, etc.
Actually, come to think of it, mixed media like those will be the last where machine translation will be able to fully take over humans, just because the text doesn't contain the entire relevant context for the job.
Keep in mind that 99% of translators are freelance. Translating something takes only a fraction of the time it takes to write the original, so publishers only have full time project managers and editors and hire translators per job as needed.
Engines like Deepl do give very good results on some single sentences. After all, they are huge databases of previous human translations, they are bound to nail it here and there. Where it falls apart is in keeping a good average and a consistent voice, so either you leave it all as messy AI sludge or you rewrite it substantially (at which point it's a glorified dictionary)
I generally agree, but don't discount the impact of platforms like Patreon and Substack. Obviously, many authors of the past had patrons, but this was generally based on status and connections - and that's to say nothing of people who can support a writing career based off inherited wealth. But there is little historical precedent for an avenue for creatives to support themselves (or at least meaningfully supplement their income, potentially leading to less non-creative work hours and thus more time to hone the craft) using a decentralized base of support.
Yeah, the arts in general in this age are reverting back to
1) the privilege of the rich and secure who can do it despite the lack of financials.
2) the ultra passionate who spend every waking moment outside of their day job engaging in the craft. Often to the detriment to their health and social obligations.
You shouldn't need to go all in on a skill just for a chance to maybe one day earn money from it.
I find myself writing more than ever. In fact I must. I have to explain in no uncertain terms what I mean. Though I often find myself in very complex technical straits where words are not enough. There's something liberating about writing a wall of text and knowing that at least something will give you feedback. It won't be annoyed. It won't be tired. It won't say "tldr?" back at you. I don't know if that will convert into a novel for me. But it must for someone.
Totally agree. LLMs are taking over everything related to intellectual work and work in general. But that doesn’t mean people will have nothing to do.
The show will go on and humanity will probably split into those who work deeply with AI and create an AI economy, where AIs develop and interact with each other (much like cloud systems do today) and a human world where people compete with other people both in art and in sports with the best receiving rewards and recognition from other humans.
There may well be some writing jobs that are safe for the reasons given in this post, but that doesn't help all the writers I know who have already lost their jobs and are struggling to find work.
It's not enough that humans can tell the difference and feel an ick, there also need to be enough organizations willing to pay money for that difference. From my vantage point, there are not. It turns out that for a ton of the writing produced by companies, the quality of the prose wasn't really "load-bearing" as Claude puts it. That writing is there to occupy a space and look professional at a glance, the same way elevator music is tolerable for the duration of an elevator ride.
I have worked in the publishing industry, translation and editing (got out to a more dependable career just in time), and what is astonishing is how even "load-bearing" text is not safe from the cost-cutting pressures. That is, even some respectable publishers for the last several years will publish a book one is expected to pay good money for, with minimal editing and proofreading compared to the old days, and if it is an originally foreign book, with machine translation and minimal post-editing.
I've got some heavy reader friends and they keep complaining that many of Spains's fantasy book translations are ridiculously bad and blatantly machine translated now: Senseless terms that require looking up the original to understand, e.g. "fall" translated as in falling when it meant autumn; proper nouns, including character names, sometimes translated sometimes not; and many sentences that mix up the grammatical gender because the machine wasn't given enough context to infer it.
But these publishers don't get punished because they keep licensing popular foreign franchises, which aren't quite fungible, so consumers don't want to miss out and thus don't vote with their wallets.
How is it possible? Even if you just put text to ChatGPT it'll translate it much better than that. And with actually decent harness, the translation should be quite good. Do they even save on tokens by choosing cheapest model or something?
It's impressive how bad it is even for machine translation, I've seen examples. You're imagining them using a good LLM and a decent harness, but I'd bet it looks more like an intern copy-pasting snippets through an LLM or even just Google Translate and cleaning it a bit.
Unfortunately the answer is no. Reading or just owning a book by a certain author can be a matter of fashion or prestige. That allows for really bad work still be published.
The problem with AI writing is that it can't add produce the texture of lived experience.
This would mean the specificity of details, thoughts and sensations as filtered through a specific individual.
There is a theory that what we perceive as beautiful in a face is that every feature, distance between the eyes, width of nose, distance between cheekbones etc is average. Note that a random face is very unlikely to have every feature be average.
With AI writing, the problem is the reverse. LLMs have captured lived experience, but it's the smooshed over experiences of millions of people.
It's not able to create a unique viewpoint which is believable. What comes out are average thoughts, sensations and details, but good writing lives in the specific.
AI's problem isn't that it can't generate unusual things. It's that it struggles to generate believable conjunctions of ordinary things.
But even this is too reductive. AIs don't have a body or a 'soul.' They don't know what it feels like to be a person, so they can't write like one.
This argument can be applied to anything AI does. AI can do bad writing, bad code, bad graphics and bad music.
But AI cannot (yet?) do the job of a skilled writer, coder, graphical artist and musician.
And it is the same problem every time, LLMs lack intention. That's essentially what the article is about, writers choose every word because they have something to convey. Something complex, deliberate, that can't fit in a simple prompt, a LLM can't get these nuances, it is all in the writer's head, so you get something generic, the information to do it better simply isn't there.
But it apply to other arts as well, ChatGPT flyers, Suno music, etc.. they all look and sound the same, because there is no intention behind them, besides all the technical issues, like inconsistent images and instruments blending into each others, the model just can't work with information it doesn't have, so it just generate something generic that looks like its training dataset.
And it is the same with code. Coding is not about the programming language, it is about expressing with precision what the machine has to do, and programming languages are really good at that, that's why they exist. LLMs let you use English instead, but it doesn't change the fact that everything has to be intentional, otherwise, the LLM will just put something that may or may not be what you want.
Most success stories of LLM coding are ports. Because someone already did most of the job with the original code, sometime even twice if we count the tests.
This is because understanding is an intrinsically personal experience. To convey it to others is to choose words, phrasing, and sentences which best express what is in one's mind.
Call it "intention", call it "understanding", call it "effective communication." The lack thereof is obvious and easily identified.
> Most success stories of LLM coding are ports. Because someone already did most of the job with the original code, sometime even twice if we count the tests.
Another way to phrase this is:
Because someone already thought about the problem and what
needs to exist in order to solve it.
Also, a huge area that AI seems to thrive in is cybersecurity, which almost by definition involves errors that the original programmer failed to think of or catch.
The thing is, at some point, it takes so much work telling the AI exactly what to do that you are better off doing the damn thing yourself.
AI can help with the execution, and for an amateur, this is important. But for real pros, the ones who actually know the appropriate style and what the layout should be, they usually have the execution nailed down. Prompting will likely drag them down, if they use AI, it is more likely to be in the form of "augmenter" tools: upscaling, content-aware fill, etc...
Again, same idea with code. Programming languages are only an issue for inexperienced coders, for those who really know what they are doing, it is usually second nature, and they may be better served writing the code themselves than trying to get a LLM to do it. Again, AI has its use: completion, analysis, etc...
That's also the reason why I believe that so many people are missing the point with generative AI taking jobs. Because they are not in the field, they only see the execution, image editing software, DAWs and programming languages look arcane, they think it is what takes the most skills, and that if you can get over it using AI, you can do just as well. The truth is that it is not the case, not by a long shot, execution is just the first step.
It is a problem for juniors however. Juniors are at the first step, they have the execution but not much more, they need to grow real skills, but how will they grow these skills if no one want them because they are at the level where AI can do most of their job.
I have a feeling the author framed this the wrong way.
my take is - the author wanted to express that there's always a demand for human prose / writing that captures the subtleties of expression, thought & ideas. which is very true. & we can already see this e.g by people opting out of LinkedIn for it's A.I driven long posts. whereas engagement is high on X where posts are likely to be human generated.
I wouldn't say writing as a job is protected - as corporations will always take shortcuts.
For now. And that is because humans are the only customers engaging with content. As agentic AI takes hold and is given credit card authority, bots will will engage. Bots will read the posts of other bots and use those when making purchase decisions for thier humans. If we are lucky, the bots will develop thier own twitter-specific language and we humans can finally wave goodbye to tweets as a concept. The bots will play the twitter game for us.
A lot of the replies are insisting AI will get better at writing with more development but I don't see it. Even if you have a mathematically perfect writing AI you still run into the same problems you would have if you handed off your writing task to someone on fiverr or something. It can't magically know what you want to say, it only has the information you gave it. A prompt complex enough where it won't get any wrong ideas has to contain as much information as the output would have.. so just write it.
> A prompt complex enough where it won't get any wrong ideas has to contain as much information as the output would have.. so just write it.
This is only if the output is fully compressed. Writing is not just about encoding the writer's ideas but also about how the reader will ingest those ideas. The writer needs to consider when to put in rests in between complex ideas to help the reader flow through the text. This suggests the LLM could be prompted by a dense complex idea to be presented with the boilerplate needed for the human mind read smoothly and without unnecessary effort.
I also believe this. Post-training LLMs with vague metrics can only be achieved with RLHF, which is not impossible, but extremely costly and difficult. Instead, companies will opt for RLVR, focusing on math and programming tasks. This pushes objectives away from writing quality; often far away. That is why older models, in my view, actually read better than newer ones. It's by design.
You can brute force it by making it try random stuff then judge itself on it. You don't have to always use an LLM's output. Sometimes you can use that plus other things to add flavor. An LLM is actually really good at judging if something is good or bad. It just has a really hard time coming up with new things. But if you had unlimited compute you can throw in some rng and whimsy and get something resembling what humans do.
I think for this argument to be true, the axiom that supports it is that the models have just as much context as they will ever have, and you cannot see being able to give them more / enough to be able to understand your perspective. That feels unlikely to be a position that doesn't change. As a society we're giving more and more context each day to this, and that makes this a valid opinion now, but one that erodes over time.
its not about dumping more and more info into the context, its about the intention. whats not in the context is just as important as what is. and i dont see how that can be automated.
also were seeing models become worse at writing as they get smarter.
Size of context is not the entire story here, it's ability to properly feed and index the context that's needed on this sort of thing. E.g. your entire slack/discord/email/github/jira/zoom meeting/coffee chat ... history is the context that you bring to the table on this sort of thing. Most of this is unindexed. Much of this will not be in the future.
> The latest models got even worse.
Which models? This is one of those things that likely has both model and domain specific aspects that impact your experience. In my experience with OpenaAI models predominantly (I previously worked there), they've improved significantly over the last 6-12 months. My experience with Claude is worse, but I haven't spent as much time getting into a mechanical sympathy there. They're still not perfect though and I have many steering docs that help avoid the biggest problems in the models I use when generating docs.
Claude writing quality got unbelievably bad with Opus 4.7, with no improvement in Fable. Opus 4.6 was fine. Im starting to see it as a security risk - my brain just can't process its word vomit, so just tell it to go on, implement whatever
No. Below is my emotional opinion based on my own experience:
Currently the models are totally helpless with plot and emotions.
They make epistemic, logistical and temporal mistakes.
But:
They are very good in finding _your_ epistemic, logistical and temporal mistakes. They are great as adversarial reviewers. They can validate grounding. They can test voices. They can build complex parallel reconstructions representing timestamped inner monologues, dialogues and events.
A process consisting of many models representing characters + states + epistemic isolation + reviewers could be great to validate or disprove ideas.
Models can write you formal models of your writing. They can run solvers against your formal models to - again - validate and disprove ideas. Literally, you could validate your writing with TLA+, SAT solvers, queries against formal ontologies.
They can build corkboards of unimaginable complexity.
A moderately good writer who plays well as a part of human-machine writing apparatus could be a competitor even for a top-tier author.
When you write with a model, you define disciplines. You write your prose. You let your model turn your prose into formal systems and verify it. You let your model criticize you. Sometimes you let your model to polish your crude language into something sleek.
So, at current state of LLM development, writer as source of ideas and governing process is safe. Writer as a vision-expansion machine is not.
I've been very hostile to LLMs in literature. I've started to write a short novel about that. As a joke I tried to use models after half of the text was done. I've applied my engineering skills. That changed both me and my opinion. Models are awesome if you know their advantages and weaknesses.
A so-so writer with a good model and a good approach to prose development could produce epic stuff.
> A so-so writer with a good model and a good approach to prose development could produce epic stuff.
Well, there have been hugos/nebulas awarded to "creative workshop" quality work before (and i mean before LLMs). But that doesn't make those books "epic".
Neither book stores nor app stores pay for quality, beyond some minimum low bar. Markets are optimized for getting product out quickly, sales and marketing, and follow on effects of succeeding with a hit. If you made a better app, game, book, or song you need to get through the noise of millions of competitors and for the most part, it's never going to happen.
In short, the expected effect of AI will be more stuff, faster, not better. Are we seeing that? I think so.
Workaday copywriting is dead (was moribund, now has been shot in the head). Prestige literary writing is zero-sum and therefore eternal, in the same way that equities trading (not formation and not business-building) is zero-sum and therefore the actual success of LLM has not given anyone a signal advantage anywhere there, because everyone else has LLM too.
Copywriting struggles with the issue that businesses know they need it, but they don't value it. It's absolutely shocking to see how poorly companies, and governments, communicate and no amount of AI/LLM usage will change it. For decades it's been neglected and relegated to "cover my ass"- writing.
The majority of AI deployment in businesses could have been avoided, if more attention had been placed on good communication. LLMs don't yield better communication or corporate writing, just more of it, because no one in business seems to understand the value or have the ability to recognize good writing.
I am not a writer but I do have to write a lot for work. The person who used to edit things for me has not been replaced, and I am expected to use AI. It is very good, but it's not all the way there. I am now left to decide whether to take the time to do it right or to go back to doing what I actually get paid to do. I wish my organization realized how important both sides of that decision tree are: we really ought to be careful about how readers perceive what we put out there, AND this isn't my job and we should have someone who's even better than me at this.
But this is an old trend with technological developments. Thanks to e-mail, most people handle much or all of their own correspondence. There used to be specialized staff for that. They were very good at it! But they were shown the door once they ceased to be strictly necessary, and now we all waste a bunch of time fiddling with Outlook instead of doing our actual work.
> It's absolutely shocking to see how poorly companies, and governments, communicate and no amount of AI/LLM usage will change it.
Thank you for saying the first part. It truly boggles the mind to see how badly most organizations communicate, even those with professional staff tasked with communications.
Regarding the second part: AI/LLM usage will lead to improvements for some people/organizations, if only for the simple reason that it removes barriers to communications. All of a sudden city hall or a third-tier supplier can quickly update their websites in multiple languages, process email communications more quickly, and empower staff who weren't good writers.
That's the part I question. You have to be a really good communicator to put yourself in the place for the reader and adjust your language to your target audience. So much of "professional" communication fails because the language assumes that the recipient is knowledgeable or even just interested. If you can't communicate effectively on your own, I doubt that you can prompt the LLM to do it for you. It's going to be very much like development, if you're a skilled developer, LLMs are a massive boost to your productivity, but if you can't code, you're just producing garbage.
AI is going to produce an increase in communication, but that is not what we need. We need better and more targeted communication. AI doesn't need to help with the targeting, we can already do that, and do it cheaper than an LLM, and again, I don't believe the majority of people are able to prompt the LLM to generate better communication, because they don't know what that looks like.
Equity trading is not zero sum. There are real businesses and assets behind the equities they represent that have actual value and produce actual money.
theoretically equity trading is about distributing stonks most optimally so that productive companies get more and less productive ones get less, improving productivity of society as a whole
Derivatives (options, futures, swaps, etc) are a zero sum market, there is always a winner and a loser necessarily because the contracts have an expiration date.
The same is not true for trading equities, they are not zero sum. There are dividends, buybacks, companies will sometimes spin off a part or parts as separate companies that you get newly issued shares of stock from (GE splitting into parts is a recent example), public companies get taken private at a premium to the market price, etc.
Futures are zero sum, but can also be a win for both sides, because what they gain is stability.
"I will buy y tons of corn from you in April for £x" - now I don't have to worry about how much my corn will cost and you don't have to worry about what your income will be.
The market has been heavily engineered and gamed to be positive sum. But sometimes I wonder could it turn out to be negative sum on certain timescales. That is that overall money would be lost on markets.
It can certainly be negative sum for a generation at a time, say 20 years. The US stock market has the good fortune of not doing that in a long time but if you look around the world for example Japan these periods are not hard to find.
Given what I have seen in recent Hollywood "writing" I would not be surprised if when correctly prompted an LLM can outdo a lot of the disposable, predictable writing in modern shows.
Every story boils down to the same predictable elements, even so-called twists are just repeats of the same old twists. Every story is boy-meets-girl, the friendly character at the start is the real villain, it was all just a dream, the power of friendship, etc.
I just asked GPT5 medium for something along those lines and got (condensed): humanity discovers a phenomenon in that the universe has started compressing causality, things start disappearing because they did not have a large enough influence upon the universe - someone's childhood hobby, a friendship that was tenuous at best and their journey is how to deal with (and ultimately mitigate) this new feature of physics.
Exactly. And it's not only about whether a specific job is safe from AI, but whether it's safe from the effects of those that aren't safe: competition from laid off people and downward wage pressure. Then there's the backpressure into the educational pipeline.
Similar, much narrower crises were managed for manufacturing jobs during the worldwide industrialization era. The solution for those displaced by trade was retraining.
Unfortunately, at the moment we have no clue of what workers should be retrained to. Not writing.
It's a pretty reliable career for hundreds if not thousands of mid-list authors who generally stick to genre fiction. It's entirely possible that AI could core out the middle of the pack as well as the bottom, which would threaten not just jobs but a wealth of institutional knowledge that those authors share among their peers and especially new up-and-comers.
If you're good at it, which extremely few people are. It's my belief that not enough good writing exists for the tech companies to train an LLM. At least not a generally applicable on.
Tech companies are training off the shadow libraries, which have the bulk of English-language human writing in book form from the twentieth century, and that traditionally published stuff with an ISBN usually got a proper editor of some kind. So, it is hard to believe LLM companies don't have enough material; it's not as if they are limited to low-quality stuff on the web.
It might come down to that they don't care (enough). They could probably do it if they had to (hire a lot of professional editors and authors to train models), but AGI seems more profitable and also a better "story" for investors than models that are good at writing.
Thing is, LLMs don't seem to be the right vehicle for AGI, but even if the nerds understand that, see above about profit perspectives and investors. These companies are under insane pressure to deliver the next super hit or be seen trying.
I have a weird background (product, development, writing + devrel). I write a lot of code and a lot of articles.
I think there is a lot of overlap in how people who write code or articles (documentation, books, etc.) use AI. On one side of the spectrum, you have people who just blindly input some prompt, accept the output and move on with their lives. You can likely predict how that is going for them (not great). On the other side of the spectrum, you have people who outright reject all AI and are continuing to plod on with how they have always done things.
In the center is a more reasonable approach that leverages AI to create without blindly accepting the output. This applies very much to writing.
The workflow that I've adopted over the past two years or so has been to leverage AI to help with the research and outline process. Once I'm happy with the structure I go and I write what I need to write.
This maps pretty closely to the code that I write. It's fine.
“Writing will remain valuable” and “writing is a safe job” are two very different claims.
AI doesn't need to write better than the best humans. It only needs to be good enough to replace a large part of the mundane paid writing that used to support writers while they developed their craft.
If I found the best human writer in the world, and had him write every article on the internet, wiping his memory between each one, his writing would start to get pretty noticeable and boring!
I agree. LLMs write really well, but probably too consistently in terms of style. That said, I did recently ask AI to do an editing pass over some writing I did, and I was surprised that it missed a tense mismatch that I had (not within a sentence, but within a paragraph).
The real thing being said here is that the author can tell good writing from bad, and assumes everyone can... I'm a visual artist and at generative art is blindingly obvious.... To me.... But not to many folks around me! Including a few artists and art adjacent folks.
It extends well beyond creative stuff too. It makes me think of Gell-Man amnesia [1].
People can tell when something they are experts in is being done poorly, but others can't, and it frustrates the experts. Then, those same people think something completely different is being done well despite what experts in that respective field say. It's like when someone from your family reads an article about your profession and then proceeds to tell you how your job works. lol
It's a really pervasive issue in society IMO. And a few prompts in someone's favorite LLM just reinforces it to people that don't know (any better|what they don't know).
Have you tested this systematically, or is it possible that you are experiencing survivorship bias? If there were any generative art pieces that you didn't notice, you would have thought that they were human-made. Therefore, all the pieces you identified were "obvious" to you. Not to mention false positives.
I imagine that they haven't done a peer-reviewed scientific study of their ability to notice generative art.
We can replace their statement by "I'm a visual artist and at generative art is often blindingly obvious.... To me....". In other words, they can notice some or most generative art but other people familiar with art can't, or at least can't as often as they do. I think what they are trying to say isn't too changed by that.
Even if it’s obvious to some at this time, as time passes artificially generated writing will become mainstream and what becomes the normal writing style. If you read IX century literature, it reads differently, the vocab is different, the sentences are built differently. It was totally normal back then. It’s not normal now. In ten years the new artificial style will become commonplace and accepted -I mean unless we artificially always make it deviate.
I think you are too close to the trees though to see the forest.
I have made digital art for 30 years and this all just sounds like what people use to say about digital art in general.
The main problem I see with generative art is not that you can tell it is generative. It is that most the art is shit. The same way if you gave a 1000 random people a blank canvas and paint, most the paintings would be shit too.
The counter example is there is a billboard that I see driving sometimes that is obviously AI generated graphics. It is so eye catching compared to any of the other billboards because most billboards are boring.
You are just puppeting the standard gate keeping bullshit to a new art form and personally I sick of reading this.
Who the fuck are you to say what art is or what art can be?
LLM prose has degraded with each model update, the models are now being RL'd into wall of texts that only makes sense to other agents. I think its going to become more obvious as we move forward that is llm generated because labs seem to only be focused on tool-calling/agentic-coding environments, as that's the only thing that drives revenue.
There may come players who focus on models that are good at writing for technical writing/docs , copyrighting ect but I think people will lean towards not using them and will rather have the "human touch" for the things that directly impact brand perception.
Keep in mind, every single AI company that is selling the idea that you don't need to hire designers and web design is "solved" have $100k retainer designers crafting their landing pages.
Yeah, this tracks with my experiments. About every 6 months or so for the past year-and-a-half I've tried using the "frontier" LLMs to as-near-as-possible autonomously write novel length stories, because I find it fascinating.
When I had Gemini 2.5 write a novel, it wasn't really objectively "good" by any stretch of the imagination, but while the prose was very purple and full of cliches and, well, bad writing I guess, it still felt ... subjectively good, at least for what it was.
Last week I did a run with GPT-5.6, and wow. On the one hand, it managed to produce 110,000 words that were "shockingly" coherent. The model was able to maintain state and plot lines and background details extremely well, much better than older models.
But I just don't like the prose. I haven't really liked _any_ prose that GPT-5.6 produces. It's significantly better at "instruction following" and keeping track of things, but, wow.
> “The sequence is consistent with their voluntary choices.” Mara enlarged the uncertainty field rather than the result. “It does not prove what happened to anyone we can’t observe. It does not prove contact did this. And it does not turn the Shard into treatment.”
GPT-5.6 in particular becomes so fixated on certain ideas like "consent" and epistemology, that by the end of the narrative, the prose and dialogue are all just "agent speech", despite the prompt/harness specifying that it's a _novel_ with narrative prose and such.
Interestingly, the model itself produces an accurate critique of its own output:
> The draft has become a *consent-centered medical, legal, and logistical procedural*. The important drift is therefore not that many events were omitted. It is that the retained events now prove a different thesis.
Which begs the question of if it would do better with a couple rounds of output -> critique -> revision. But I think I've had enough LLM prose for a bit...
The entire output of an LLM is also part of its input. Appending to long stretches of LLM-generated text, it will continue to get more and more robotic as the style of the input gets replicated and enhanced in the output.
It's possible to work around by generating short passages at a time with carefully constructed setup. But it's a real pain.
It's that, yeah, but coming from multiple orders of abstraction.
In this case, part of the experiment was to see what "oh-my-pi", a "fat and feature rich" LLM harness, could do when coupled with modern GPT, given a 6k~ word overview of a story, and told to come up with a plan to write/review/audit it, making use of subagents and all the fun new groovy LLMisms...
Part of the problem was just "it was basing its style off the last scene/chapter", but part of it was also that its instructions were constantly being "compressed" through repeated compactions. Even with the use of subagents, the "top level" agent's prompt was getting muddied, and in the "review" phase, it began to focus more and more on creating increasingly complex ledgers.
You can see this happen in the "plan" files it created for each chapter, looking at word count:
So it wasn't just that the prose was being based on an increasingly compressed "style" of the prior context window, but the planning for writing each scene was, itself, becoming fixated on the "continuity error correction" process itself, to the point where by the end, it had mostly forgotten about the prose part, and was completely fixated on ensuring maximum state continuity.
This could definitely be fixed, but honestly, I've about had my fill of the "autonomous writing agent" goal. The idea was to make a model that could generate sufficiently interesting stories based on "vague premises" for my personal entertainment, but, "surprise", getting LLMs to actually produce both "new" and "coherent" content beyond what you specify is _hard_.
It seems like you really do need to just stay "in-the-loop" with every scene, and constantly provide correction/feedback, to correct the "semantic drift".
Or, gasp, I could just try writing things by hand again... :-)
Claude was really far ahead of GPT in writing from 2-4, but the later models have started to get an overly distinctive style, wheras these days GPT tends to be coherent but fairly concise and dry.
Claude is the one model family I've not really used. Which yeah, feels like a backwards thing to say in a world where seemingly everyone using LLMs is using Claude Code.
I've used some Opus 4.5/4.6 via Antigravity and Sonnet by the web chat. I'm torn because as far as LLMs go, it does feel more ... "literate".
But maybe too literate, judging by how many people are complaining about "Claudeisms". I suspect Claude would be just as susceptible, if not more, to the sort of ... "moralizing" that GPT seems to gravitate towards (for lack of a better term).
I don't think writing is an "AI-complete" problem. It's just that the models are inefficient at it right now, due to training and/or architecture. Good writing requires thinking, reflection, and doing multiple passes. This currently translates to using a high reasoning effort and burning lots of tokens, but big AI labs like Anthropic are struggling with handling the load, so they're doing the exact opposite and finding whatever cheap trick they can to reduce token usage. One such trick is condensing ideas into as few tokens as possible, which in my opinion is one of the reasons Claude sucks so much at writing.
That's because it's cheaper to automate than pay a software engineer. The economics has to make sense. Writing is tricky. Prose quality isn't there yet for many genres. You can tell AI written prose apart. Maybe it won't matter. Maybe there is demand for it even. Or Maybe we will create genre specific finetunes if there is enough demand for it.
Yeah I had GPT5.6 Sol act as an editor on my new chapter and it did find decent issues with PoV and narrative distance but it was so bad at prose. The best way I can describe it is, it was too clean. Often times humans write about things unsaid and left for reader to interpret or deliberate awkward sentences to tease character psychology. The AI just straight up corrects these without second thought. This will likely be the case with general purpose models unless we see some genre specific fine tunes.
If we're doing book recommendations, I quite recommend the essays in https://academic.oup.com/mit-press-scholarship-online/book/1... and particularly the philosophical discussion there of whether "robot sex" is an oxymoron, as a robot almost by definition cannot dissent, so it might only be "robot-assisted masturbation"
If you thought all of the money being poured into data centers was excessive, wait until people can order up a proto-sentient sex bot to their front door. Oh boy.
I think there is good chance that AIs have already replaced parts of the certain aspects of industry. And with some tuning could probably replace lot more of at least virtual parts...
I think writing will go the way of software development. Really good writers and devs will probably thrive with AI but most of us do fairly mundane, repetitive work that will get done by AI.
I remember, when I used ChatGPT the first time to write some E-mails and other documents, I suddenly realized that I sound exactly like corporate communications or most tech writers. That made me realize that these guys are also working off templates and basically produce variations of the same thing. Same as developers do.
AI writing is soooo annoying. It's always way too long and way too "serious" for what ultimately is almost always a really dumb point. There is no calibration to the seriousness of the topic or thesis against the actual length of the writing. It's insufferably pompous. Will it get better? Probably.
Arriving ten hours after this was posted, I am a tad surprised that nobody seems to have popped up to suggest that the article ought to have been titled, "The job safest from AI may be writing".
Maybe eventually, but not now. Our AI-crazed org is doing everything possible to have us automate our jobs out and has laid off anyone who isn't hitting their targets to do so. Our CEO point-blank said he wants our department cut by end of the FY. Same org just acquihired dozens of senior SWEs.
"As AI slop saturates the web, with the same token (pardon my pun), the economic value shifts entirely to an authentic human voice."
In that respect, I'd be more than happy to read more direct quotes from Gemini and then read what a live hacker's take is on it. Slop provides the building blocks, and humans are the alchemists. Need both.
I appreciate what this piece is trying to do directionally but it shares a logical flaw with lots of other pieces about AI and knowledge work. It assumes that in order to disrupt a particular field, AI needs to print serviceable work unassisted; that it's "vibe-shipping".
AI today is most effective when it's not vibing, but rather copiloting a skilled operator. I don't necessarily want my agent to build an entire system for me, even if it's ultimately the author of almost every line of code; I'm actively making decisions throughout. There's obviously a spectrum here but at most points on the spectrum the amount of assistance available is still a step change in the economics.
So too with writing.
The first rule of accelerating writing with AI is that you're not allowed to use a single word the AI suggests. Even if what the model comes up with is great, better than what you could have done, as soon as the AI suggests it it's poisoned. At least with current models, readers can detect LLM prose in the parts per trillion, and as soon as they do you've lost them.
The second rule of writing with AI is that AI encouragement is toxic. A structural consequence of RL is that models are exquisitely tuned to generate responses that make their users perceive value. We recognize this in a gross sense in "sycophancy", but the problem recurs fractally in at finer-grained levels, where stuff like "this part is really strong" will subtly allow the model to set a course for your writing and you'll confidently ship crap.
With those two rules in mind, models are incredibly valuable for writing, more valuable in my experience than the professional copywriters I've worked with. The trick is to get them to make suggestions at a higher level than just writing alternatives:
* Do the sentences in these paragraphs end with the new idea or information?
* Are the real actors in each sentence the grammatical subjects?
* From paragraph to paragraph is there a clear flow of topics, or are things jumping around?
* Is this piece crudded up with metadiscourse like "it's important to note"?
I've had a stack of notecards for ages that I took down from Joseph Williams "Style: Towards Clarity And Grace", the most programmer-brained writing book ever written, I love it very much. For the past year or so I've been feeding them through GPT and Claude one by one, and it's drastically increased the speed at which I can knock out a completed piece.
I think it's pretty hard to argue that AI isn't going to have an impact on the writing profession. It's just not the most obvious impact everyone assumes it will have, where it, like, writes whole op-eds or whatever. At least not yet.
Writing has an audience, purpose, constraints, and desired outcome. Creating tests for writing is more difficult than creating a unit test, but it's not undefinable. It's just engineering.
First, I actually had the author of this post as my distributed systems professor back in college. He's not just a super smart dude on a technical level, but he's probably the only CS professor I've ever had who seems to appreciate writing in itself (which may be evident from his blog). I think the combination of these two qualities puts him in a good spot to make the claims he is.
In terms of what he's presenting here, I'll say this: his line of logic highlighting the "dual-mind problem" of writing seems, to me, like the most compelling aspect of this argument. When I write something (especially to a specific audience, including an individual), I must practice cognitive empathy if I want what I write to be consumed properly and effectively. This is a cold way of putting it, but even in text message responses which span a few words towards family or a friend, I can put quite a bit of thought into how it will be received, how they will read it, interpret it, etc. My cadence in just texting, alone, can shift dramatically from one message to a next based on who I'm sending it to, what I'm trying to convey, etc.
All of that is fine and not hard to understand, but I suppose the interesting part is this: when I'm just trying to get information across (i.e., instructions, directions, etc.), my messages will resemble something written by an AI. It's not enjoyable to consume, but that's not the point. But as soon as I want to add a bit of "fun" to a message, the task I'm performing is completely different. It's no longer just an exchange of information, but it's an attempt to invoke specific feelings, visualizations, memories, etc., in the other person/people.
I do have a hard time imagining what it will take for AIs to be able to do THAT effectively. I think they're fine at conveying information, but I think people are already becoming very aware that conveying information, in itself, is not enough to be effective. These LLMs don't "care" to "entertain" you with what they're writing at you about. They're just spitting out the mathematically derived facts with, seemingly, no meaningful ability to invoke deeper thoughts in their audiences' minds. And that might not seem particularly important outside the context of writing fiction, etc., but I think it's actually pretty critical even in "dry" settings, like explaining code, because the thoughts, feelings, emotions, etc., invoked through reading IS the output of reading (even if it doesn't seem that way).
Perhaps AI will be able to do this more effectively if they're trained on direct brain signals or something. Like, train the AI not just to convey accurate information, but also encourage it to do so in a way which stimulates different areas of the brain with different intensities based on the premise that, doing so, is actually what people are getting out of that text.
> The stuck-in-slop state of LLM writing is not from lack of trying. AI labs already tried hard on improving prose and hit a wall. LLM giants would have loved to ship better writing capability to conquer marketing, copywriting, and publishing at zero marginal cost.
This doesn't ring true to me.
LLM prose can be greatly improved with the right prompts.
Now, prose with the right prompts may well still not be as good as a good human writer would write – but it is a lot better than what LLMs produce by default.
If the model can perform better with the right prompts, it suggest they haven't actually done everything they could in the post-training to maximise writing quality.
I was thinking about this yesterday, while going through some slop documentation generated by deepseek. If you compare claude fable 5, side by side with the deepseek, the differences in prose are glaring. It's not even close. Deepseek prose reads like fragmented shorthand, where claude fable 5 comes fairly close to human, certainly not superior to human quality writing.
I watched a podcast with a cognitive scientist and one of main contributors to the theory of linguistic relativity, Lera Boroditsky.
She said something to the effect that, "in this very moment, we are speaking in ways that were never spoken before. We are saying things that no other person has said before...."
Language models are not sample efficient and cannot adapt to evolving language, unless it's documented in large amounts of examples.
So whatever isn't documented, whatever isn't in the training dataset or the rag corpus, the model will always be incredibly different in expression from humans.
I recently have been trying out frontier models on writing. Just for laughs, to bring an idea I have into the world acting more as director than author. I have a multi agent setup and am getting different models to argue about rating the story for clarity and whether the plot is complete.
I agree with the article. Even the frontier models call out that all the characters tend to sound the same. It also started at some point making huge changes to the core premise, and also adding characters willy nilly. The issues it called out with the plot (the ones it actually consulted me on) also made me realize how terrible of a writer I actually am.
Overall, it has been an interesting experience. I look forward to reading my own book!
I understand a lot better now why people are bemoaning KDP being filled with absolute garbage AI slop.
I disagree, and not because I think AI can write as good as a good writer does, but because in 90%+ of the cases, _companies won't care_.
I did translation as a side job for about 20 years (helped me keep my language skills sharp, and some side income was welcome). I was good at it. But translation was one of the first professions to fall to AI (even before LLMs but especially since), and I saw increasingly that companies were willing to accept the clearly lower-quality work if it meant saving money. A couple of years ago I stopped doing any translation work altogether.
Sure, whoever is hired to translate Murakami's latest novel will be a human, a good writer first and foremost. But that translation work is a tiny fraction of the overall translation work contracted by companies.
The same thing is going to happen with writing. It won't disappear, but demand will drop by >80%.
> ... and model checkers can instantly catch errors.
Lol. If that were true software would've been a lot better historically... Model checkers don't scale to 90% of the software we write. Typically you have to 1) heavily abstract the program and 2) put it in some sort of harness to specify how you want to model the outside world (which will always fall short of practice). And probably 10 other workarounds since most model checkers are Research Grade Software™. Not saying they're not tremendously useful, but that bullet doesn't hold up at all.
"Humans are going to be significantly better at writing things intended for other humans to consume" is completely at odds with "And organizations care enough about the gap, and are willing to pay for the difference".
Go on to Ao3 and a popular fandom, say like A Song of Ice and Fire: 9 out of 10 fanfictions written in the past two years stink to high heavens of AI. 9 out of 10. From some that are obvious AI to even normies, to some that slip one or two ai tropes after a couple of chapters, because the author got lazy and stopped editing carefully.
It's sad, infuriating, discouraging. A deluge of slop drowning the last embers of authentic human creativity.
This feels premature. The thing about the LLMs people use today is that it's a handful of super expensively trained models serving 1000s of use cases ranging from frontier math to recipe planning.
We get to learn the foibles and language of Claude and ChatGPT as a result. The slop is almost detectable if you provide no steering prompts about story structure, narrative structure, or stylistic cues. And most writers are not finetuning the weights to their LLMs explicitly.
If you invest time into doing all that (not really trivial stuff), the results will be better.
The safest jobs are the ones that can't be done behind a screen. I'm a technician and there's no robot that could replace me and there won't be for a long time.
You can't make 1000+ trash robots just to produce one that works okay sometimes and somehow make the economics work from that, so I very much doubt it.
Edited to say "I think". I think that because I have been looking hard for developments in humanoid robotics generalization and new impressive demos and capabilities keep popping up all the time.
For example the recent Skild AI fairly general imitation capability. It really is a huge amount of effort on increasing the generality of humanoid robots going on now and a lot of demonstrations coming out showing off progress.
With the current trajectory there is no reason to think it will stop soon. There is a lot of motivation to innovate and train and it is pushing things forward.
If you're not looking for evidence of improvement, or rather looking for failure cases, then you're not going to see that trajectory.
And physical damage is lot more expensive to repair in vast majority of use cases. Have fun when your humanoid window washing robot punches through your window. Or any other such cases.
Except that in the worst-case AI scenarios where a huge portion of the professionals who work "behind a screen" lose their jobs, the entire labor market will have to deal with that.
There's so much focus in the US on vocations but Americans have no idea how little mechanics, plumbers, etc. in other countries make. There are a variety of reasons for this but one of the biggest is that you can't charge $200/hour to fix a leak when $200 represents a double-digit percentage of a worker's monthly salary.
If white collar workers get laid off tradesmen are not going to be able to charge much any more either. Not to mention the price pressure that will accompany the increased supply of workers.
I somewhat agree. I believe that anything and everything will eventually be "taken over" by AI. The question is when. Writing and design that isn't slop will be two of the last pieces to go. Anything that involves math will be the first.
The reason LLMs use the same cadence and cliches is that their RLHF does not emphasize writing well, the way it emphasizes, say, coding. And the reason for that should be pretty obvious: coding is where the big money is at.
So the author may be correct, but for a different reason: unless writing starts being very valuable as a profession, it's unlikely the labs will spend significant resources making their AI models better at it.
No, they use "the same cadence and cliches" because they inflate a short and ambiguous prompt into long and specific prose by making statistical assumptions about what best fills in the gaps. It's not a training problem, it's an information theory problem, and it's not really surmountable.
Any given model will always have some distinct implicit voice that its biased towards for that infill content, and so a popular model will always become exhaustingly common, painfully familiar, and cliche. Users can use more elaborate prompts that shift the voice away from the most normative and towards some other nodes, but they need to put in special effort for that, and what people-at-scale specifically want from these tools is to put in very little effort, so we can expect that overwhelming number of casual and naive users will always be generating cliche slop with them.
Code escapes this problem not because of training but because it specifically benefits from cliche (boilerplate, patterns, etc) and so an model whose code "voice" reflects your own taste as a coder (or your toolchain's taste as a vibecoder) is going to feel like productive output rather than slop. But it's still cliche.
> No, they use "the same cadence and cliches" because they inflate a short and ambiguous prompt into long and specific prose by making statistical assumptions
Even if LLM output has to largely follow some statistical rules, yet, first of all, some amount of randomness is normally injected during token generation, and, secondly same true for human speech.
> about what best fills in the gaps. It's not a training problem, it's an information theory problem, and it's not really surmountable.
This is not true, as LLM has internal knowledge storet in its weight. Unless you force it to produce 2000 words doc out of 3 word prompt, you would end up adding some sense information.
>Users can use more elaborate prompts that shift the voice away from the most normative and towards some other nodes, but they need to put in special effort for that, and what people-at-scale specifically want from these tools is to put in very little effort, so we can expect that overwhelming number of casual and naive users will always be generating cliche slop with them.
True, here I agree with you. But using finetuned or simply less popular models like Kimi, Hy etc. should take care of that.
No. Reinforcement Learning is doing a lot here. Anyone who played with these models before the Davinci intstruct-tuning (completion) era can tell you the same. In some ways, SOTA models have gotten better at writing, but the neuroticism of instruct-tuning has still not been resolved.
"The reason LLMs use the same cadence and cliches is that their RLHF does not emphasize writing well, the way it emphasizes, say, coding. And the reason for that should be pretty obvious: coding is where the big money is at."
Software can be checked for being 'written well' by compilers / linters etc. There is no equivalent for well-written natural prose. Spelling and grammar checkers haven't a clue about prose semantics.
They just haven't got to it yet. It's not a high priority but eventually they will make a solid effort on easier style guidance and more humanization. They will probably drop the most worn ou constructs like "not X, not Y, but Z" that have become tells and target a lot of known issues in reinforcement training.
It's slightly weird how confident writers are that it won't get improved.
What's interesting is that we still read AI's output and that's still the primary interface between a human and the AI. Implicitly that means that writing is important and I think we're already starting to see this with the complaints around how Claude "talks"
Claude "talk" is particularly annoying if you're using Claude to perform a job function (write code, analyze a spreadsheet, etc.) but average people reading an ad, Facebook post, etc. don't really care from what I can tell. In fact, I think most of them have no idea what Claude "talk" is, even if some of them notice common patterns that have emerged recently.
I think it's likely to be lawyers who are the safest. Lawyers are the ones primarily in charge of the government, and I think it's likely that as soon as they feel like the legal profession is threatened, they will pass laws making it illegal to use AI for legal issues. It's not all sunshine and roses (paralegals are probably hosed), but the lawyers seem like they'll do ok.
Unless the model watermarks text, it will become increasingly difficult to tell the difference between human legalese and AI legalese. Especially if they make it illegal, that will create a strong incentive to find-tune to imitate human lawyers more convincingly.
There's already been one case where the prompts were allowed to as a subject of discovery, and lying about it is perjury. Lawyers really are probably safe.
Eh, I think we're going to run into the same issues we're seeing in software. That is, junior and lower status positions (so SDETs in software, paralegals in law) are rapidly being downsized, and that's going to have a big issue in the future where there isn't the same talent funnel. And then in the future those that work their way up will basically be dependent on working with AI.
It is great time to start business as downsized paralegal. While ago there was an app that objected parking ticket fines in NYC. Anyone can find similar niche to specialise in, and just produce boring repetitive law work using LLM templates and mostly automated work.
> I think it's likely to be lawyers who are the safest. Lawyers are the ones primarily in charge of the government
Given what seems like an increasingly inevitable deprecation of these outdated, lumbering nation-states, it seems to me that these two assertions are mutually exclusive.
People seem to have very short memories. LLMs are easily capable of good writing. The fact that they are loaded up with obvious "tells" now is quite a recent development.
One year ago, the writer Mark Lawrence posted an article [0] (featured on HN) describing how he generated 4 pieces of flash fiction with ChatGPT, asked 4 published authors (with a combined book sales of $15M) to also write a flash fiction piece, and then did a blind test asking respondents, for reach of the 8 pieces, (a) whether they thought it was written by a person or not and (b) an overall quality score. The result was that people (who were avid readers and skewed anti-AI) were no better than a coin toss at determining whether the piece was written by ChatGPT, and slightly preferred ChatGPT-written writing overall.
This is something that constantly gets missed by HN in discussions about AI in the arts. There's the quality of the output and the essence behind the art itself. Humans want both [1][2] but these models only provide an output. Could ChatGPT output the next IT or Lord of the Rings, probably.
However, once it's understood that it wasn't written by a human it loses all value. I use this quote by Kane Parsons a lot but I think it summarizes why humans like art.
> If I see an element of the environment has been, you know they used generative fill or whatever to change something about the scene, even if--. It just shuts the part of my brain that wants to know more about that world and wants to look for details, 'cause I would assume if they're willing to make an arbitrary choice there, they can make an arbitrary choice with literally anything.
Specifically, people like dissecting art whether its writing, a movie or an image. It could be seeing the Chekov's Gun finally fire, and going back through the pages to see the author wrote it in there earlier and you missed it. It could be speculating about the unopened questions the author deliberately left there. This is how a lot of people at least in my circles engage with art and AI provides none of that. Even if the author specifically prompts that into it, those arbitrary choices are going to be made in certain spots that just disinterests me.
> However, once it's understood that it wasn't written by a human it loses all value.
Hmm...No. I actually enjoy generating short stories for myself - after light editing they are great; they are enjoyable irrespective of the provenance.
I largely agree with your take on the subjective value of writing that I happen to know was generated by a machine, but that is not at all what I'm talking about here. I'm responding to the specific claim in TFA that LLMs produce low-quality output, detectable independently of whether the reader knows how it was produced:
> I am appalled at the absolute shit [LLMs] spew as prose. They always follow the same robotic cadence and cliches, and sprinkle the same tired vocabulary all around.
This claim is demonstrably untrue, and central to the author's overall thesis.
What I find fascinating, in a let's say sociological manner is how we've been thoroughly brainwashed to accept things like the complete destruction of middle class by their replacement with AI.
Everybody treats this as something innevitable, like the movement of the tectonical plates.
Everybody knows this is being paid with the giant transfer of wealth from the working classes towards the asset owning class via profilgate government expense, ZIRP and QE policies from the FED and the covid so-called stymulus. This created, via Cantillon effect, inflation for consumers and a bizarre and absolutely abnormal deluge of capital for our financial system masters that let they play God and make us, the plebe, obsolete. And yet we don't question anything.
We are accepting as inevitable and uncontrollable something that only looks to be like that.
I know, we all have our lives, out of the confortable anonymate here, I am a enthusiastic AI Booster. Out of some true zealots, we are all hedging our bets to stay on the right side of the city walls of the new techno-feudalism. But, realistically, how many of us the Musks and Thiels will need after they obsolete most of us, no matter how much skills and knowledge anyone of us may have? Just betting to be on the side of the oppresors is a very long bet my friends.
Nothing about it is inevitable, nothing about it requires, as if it was a law of nature, to be unregulated and the people be damned.
Not so many centuries ago, we had peasants riotting and guillotining their opressors, we had proletarian revolutions. They didn't have encryption, the internet, cell phones, digital radio modes that allow you to talk to some other partisan group accross the globe using a cheap chinese SDR and a simple dipole hang between trees.
Things can be different. We still can force some democracy down their throats.
I don't think there's any chance people will be employed to sell the output of their brains in 20 years, outside of "athletes", or whatever we end up calling people who think for sport. We mostly succeeded at doing this with muscles, and today there's no market for hauling rocks out of quarries manually. We're in the process of doing that for brains too.
AI will outdo people in all practical uses. We're already there for debugging and getting very close for coding, and we're in the middle of the largest investment in human history to expand that to everything else.
If we do a good job of alignment, AI will treat people like those cats "in charge" of train stations in Japan: our every need will be accommodated, but we won't be controlling things we don't understand.
Of course. But they're going to be machines. I don't think there's magic that makes the human brain impossible to automate. And engineering is going to lead to faster improvements than evolution could.
People will say that until it doesn't. We're currently in the largest research investment in history for this, and have made a crazy amount of progress since even five years ago.
I'm not cocky enough to bet against it. Especially since now AI is solving frontier math problems, debugging better than humans, and writing most of the content posted to Hacker News.
If that's how you feel, you'd better get started -- we've already got AI writing most of the posts on this site, handling frontier maths, writing most of the commercial code at most startups, and its capabilities are growing real fast.
It doesn't matter if LLMs can't hold a candle to some of the finest human writing. What matters is that before you become a world-renowned writer, you need to pay your bills, often for a long time. And that's what LLMs take away. All the mundane literature-adjacent works: journalism, translations, technical writing / corporate comms, copyediting.
The same goes for many other forms of art. A budding painter or a musician could support themselves off commissioned / commercial work while working on their grand opus... but now, the customers just prompt gen AI.
The first careers I witnessed being negatively impacted by AI coming onto the scene .. just a couple years ago .. was tech writers. Source: my personal network of tech writer colleagues.
Was a beautiful house, which I briefly lived in during scatterbrained renovations (never met the new owner).
The bank forclosed on his house ~late2024/early2025... anybody can now just use an LLM to translate (in real time), on sub-$300 hardware (offline, open-source models run via e.g: Whisper[GUI]).
Either your colleagues are facing stiffer competition off-shore (not LLM driven) or their jobs were more about organizing, checking and sorting topics and data (LLMs can take that away)
>A budding painter or a musician could support themselves off commissioned / commercial work while working on their grand opus... but now, the customers just prompt gen AI.
Are you joking? The only people I see using generative AI are either companies, shitposters, and spammers. Who are all these people who have stopped paying commissions and moved over to prompting AIs?
Especially on forums like HN, the AI isn't generating the person's views - it's helping them say that what they already mean more clearly in English.
>a limited amount of diplomatic communiques
With zero data to support it, I'd bet good money the vast majority of semi-professional translators (that is, those not employed by publishers but still making some money off of their work) work on fiction, translating comics, subtitles, etc.
Actually, come to think of it, mixed media like those will be the last where machine translation will be able to fully take over humans, just because the text doesn't contain the entire relevant context for the job.
Full time professional, 20 years of experience.
Keep in mind that 99% of translators are freelance. Translating something takes only a fraction of the time it takes to write the original, so publishers only have full time project managers and editors and hire translators per job as needed.
Engines like Deepl do give very good results on some single sentences. After all, they are huge databases of previous human translations, they are bound to nail it here and there. Where it falls apart is in keeping a good average and a consistent voice, so either you leave it all as messy AI sludge or you rewrite it substantially (at which point it's a glorified dictionary)
Harness issue, just explicitly tell them to maintain a vocabulary as they go for proper nouns and novel terms.
The replacement is not better, but it’s good enough for the undiscerning people who prefer the cost pr convenience of LLM-generated content.
In any case, you find yourself competing with oligopolies that can insert their bad LLM output in front of your good work. It’s not a fair fight.
1) the privilege of the rich and secure who can do it despite the lack of financials.
2) the ultra passionate who spend every waking moment outside of their day job engaging in the craft. Often to the detriment to their health and social obligations.
You shouldn't need to go all in on a skill just for a chance to maybe one day earn money from it.
The show will go on and humanity will probably split into those who work deeply with AI and create an AI economy, where AIs develop and interact with each other (much like cloud systems do today) and a human world where people compete with other people both in art and in sports with the best receiving rewards and recognition from other humans.
It's not enough that humans can tell the difference and feel an ick, there also need to be enough organizations willing to pay money for that difference. From my vantage point, there are not. It turns out that for a ton of the writing produced by companies, the quality of the prose wasn't really "load-bearing" as Claude puts it. That writing is there to occupy a space and look professional at a glance, the same way elevator music is tolerable for the duration of an elevator ride.
But these publishers don't get punished because they keep licensing popular foreign franchises, which aren't quite fungible, so consumers don't want to miss out and thus don't vote with their wallets.
This would mean the specificity of details, thoughts and sensations as filtered through a specific individual.
There is a theory that what we perceive as beautiful in a face is that every feature, distance between the eyes, width of nose, distance between cheekbones etc is average. Note that a random face is very unlikely to have every feature be average.
With AI writing, the problem is the reverse. LLMs have captured lived experience, but it's the smooshed over experiences of millions of people.
It's not able to create a unique viewpoint which is believable. What comes out are average thoughts, sensations and details, but good writing lives in the specific.
AI's problem isn't that it can't generate unusual things. It's that it struggles to generate believable conjunctions of ordinary things.
But even this is too reductive. AIs don't have a body or a 'soul.' They don't know what it feels like to be a person, so they can't write like one.
But AI cannot (yet?) do the job of a skilled writer, coder, graphical artist and musician.
And it is the same problem every time, LLMs lack intention. That's essentially what the article is about, writers choose every word because they have something to convey. Something complex, deliberate, that can't fit in a simple prompt, a LLM can't get these nuances, it is all in the writer's head, so you get something generic, the information to do it better simply isn't there.
But it apply to other arts as well, ChatGPT flyers, Suno music, etc.. they all look and sound the same, because there is no intention behind them, besides all the technical issues, like inconsistent images and instruments blending into each others, the model just can't work with information it doesn't have, so it just generate something generic that looks like its training dataset.
And it is the same with code. Coding is not about the programming language, it is about expressing with precision what the machine has to do, and programming languages are really good at that, that's why they exist. LLMs let you use English instead, but it doesn't change the fact that everything has to be intentional, otherwise, the LLM will just put something that may or may not be what you want.
Most success stories of LLM coding are ports. Because someone already did most of the job with the original code, sometime even twice if we count the tests.
Call it "intention", call it "understanding", call it "effective communication." The lack thereof is obvious and easily identified.
> Most success stories of LLM coding are ports. Because someone already did most of the job with the original code, sometime even twice if we count the tests.
Another way to phrase this is:
"Give me a poster for my band playing a gig"
Versus
"Give me a poster for my band playing a gig, it should have x y and z. Use a x' artistic style and include elements of y'. The layout should be z'..."
You ask for the default, you get the default.
AI can help with the execution, and for an amateur, this is important. But for real pros, the ones who actually know the appropriate style and what the layout should be, they usually have the execution nailed down. Prompting will likely drag them down, if they use AI, it is more likely to be in the form of "augmenter" tools: upscaling, content-aware fill, etc...
Again, same idea with code. Programming languages are only an issue for inexperienced coders, for those who really know what they are doing, it is usually second nature, and they may be better served writing the code themselves than trying to get a LLM to do it. Again, AI has its use: completion, analysis, etc...
That's also the reason why I believe that so many people are missing the point with generative AI taking jobs. Because they are not in the field, they only see the execution, image editing software, DAWs and programming languages look arcane, they think it is what takes the most skills, and that if you can get over it using AI, you can do just as well. The truth is that it is not the case, not by a long shot, execution is just the first step.
It is a problem for juniors however. Juniors are at the first step, they have the execution but not much more, they need to grow real skills, but how will they grow these skills if no one want them because they are at the level where AI can do most of their job.
my take is - the author wanted to express that there's always a demand for human prose / writing that captures the subtleties of expression, thought & ideas. which is very true. & we can already see this e.g by people opting out of LinkedIn for it's A.I driven long posts. whereas engagement is high on X where posts are likely to be human generated.
I wouldn't say writing as a job is protected - as corporations will always take shortcuts.
This is only if the output is fully compressed. Writing is not just about encoding the writer's ideas but also about how the reader will ingest those ideas. The writer needs to consider when to put in rests in between complex ideas to help the reader flow through the text. This suggests the LLM could be prompted by a dense complex idea to be presented with the boilerplate needed for the human mind read smoothly and without unnecessary effort.
Doesn’t stop people trying.
I think for this argument to be true, the axiom that supports it is that the models have just as much context as they will ever have, and you cannot see being able to give them more / enough to be able to understand your perspective. That feels unlikely to be a position that doesn't change. As a society we're giving more and more context each day to this, and that makes this a valid opinion now, but one that erodes over time.
also were seeing models become worse at writing as they get smarter.
> The latest models got even worse.
Which models? This is one of those things that likely has both model and domain specific aspects that impact your experience. In my experience with OpenaAI models predominantly (I previously worked there), they've improved significantly over the last 6-12 months. My experience with Claude is worse, but I haven't spent as much time getting into a mechanical sympathy there. They're still not perfect though and I have many steering docs that help avoid the biggest problems in the models I use when generating docs.
Currently the models are totally helpless with plot and emotions.
They make epistemic, logistical and temporal mistakes.
But:
They are very good in finding _your_ epistemic, logistical and temporal mistakes. They are great as adversarial reviewers. They can validate grounding. They can test voices. They can build complex parallel reconstructions representing timestamped inner monologues, dialogues and events.
A process consisting of many models representing characters + states + epistemic isolation + reviewers could be great to validate or disprove ideas.
Models can write you formal models of your writing. They can run solvers against your formal models to - again - validate and disprove ideas. Literally, you could validate your writing with TLA+, SAT solvers, queries against formal ontologies.
They can build corkboards of unimaginable complexity.
A moderately good writer who plays well as a part of human-machine writing apparatus could be a competitor even for a top-tier author.
When you write with a model, you define disciplines. You write your prose. You let your model turn your prose into formal systems and verify it. You let your model criticize you. Sometimes you let your model to polish your crude language into something sleek.
So, at current state of LLM development, writer as source of ideas and governing process is safe. Writer as a vision-expansion machine is not.
I've been very hostile to LLMs in literature. I've started to write a short novel about that. As a joke I tried to use models after half of the text was done. I've applied my engineering skills. That changed both me and my opinion. Models are awesome if you know their advantages and weaknesses.
A so-so writer with a good model and a good approach to prose development could produce epic stuff.
Well, there have been hugos/nebulas awarded to "creative workshop" quality work before (and i mean before LLMs). But that doesn't make those books "epic".
This is similar to observations that effect of AI on quality of apps in stores is mostly non-existent.
In short, the expected effect of AI will be more stuff, faster, not better. Are we seeing that? I think so.
The majority of AI deployment in businesses could have been avoided, if more attention had been placed on good communication. LLMs don't yield better communication or corporate writing, just more of it, because no one in business seems to understand the value or have the ability to recognize good writing.
But this is an old trend with technological developments. Thanks to e-mail, most people handle much or all of their own correspondence. There used to be specialized staff for that. They were very good at it! But they were shown the door once they ceased to be strictly necessary, and now we all waste a bunch of time fiddling with Outlook instead of doing our actual work.
I assume you meant "now" instead of "not": "The person who used to edit things for me has now been replaced,"
Thank you for saying the first part. It truly boggles the mind to see how badly most organizations communicate, even those with professional staff tasked with communications.
Regarding the second part: AI/LLM usage will lead to improvements for some people/organizations, if only for the simple reason that it removes barriers to communications. All of a sudden city hall or a third-tier supplier can quickly update their websites in multiple languages, process email communications more quickly, and empower staff who weren't good writers.
That's the part I question. You have to be a really good communicator to put yourself in the place for the reader and adjust your language to your target audience. So much of "professional" communication fails because the language assumes that the recipient is knowledgeable or even just interested. If you can't communicate effectively on your own, I doubt that you can prompt the LLM to do it for you. It's going to be very much like development, if you're a skilled developer, LLMs are a massive boost to your productivity, but if you can't code, you're just producing garbage.
AI is going to produce an increase in communication, but that is not what we need. We need better and more targeted communication. AI doesn't need to help with the targeting, we can already do that, and do it cheaper than an LLM, and again, I don't believe the majority of people are able to prompt the LLM to generate better communication, because they don't know what that looks like.
The same is not true for trading equities, they are not zero sum. There are dividends, buybacks, companies will sometimes spin off a part or parts as separate companies that you get newly issued shares of stock from (GE splitting into parts is a recent example), public companies get taken private at a premium to the market price, etc.
"I will buy y tons of corn from you in April for £x" - now I don't have to worry about how much my corn will cost and you don't have to worry about what your income will be.
These folks compete in a reasonably zero-sum game.
Every story boils down to the same predictable elements, even so-called twists are just repeats of the same old twists. Every story is boy-meets-girl, the friendly character at the start is the real villain, it was all just a dream, the power of friendship, etc.
I just asked GPT5 medium for something along those lines and got (condensed): humanity discovers a phenomenon in that the universe has started compressing causality, things start disappearing because they did not have a large enough influence upon the universe - someone's childhood hobby, a friendship that was tenuous at best and their journey is how to deal with (and ultimately mitigate) this new feature of physics.
From what I heard it's a bloodbath at the bottom.
Similar, much narrower crises were managed for manufacturing jobs during the worldwide industrialization era. The solution for those displaced by trade was retraining.
Unfortunately, at the moment we have no clue of what workers should be retrained to. Not writing.
Thing is, LLMs don't seem to be the right vehicle for AGI, but even if the nerds understand that, see above about profit perspectives and investors. These companies are under insane pressure to deliver the next super hit or be seen trying.
I have a weird background (product, development, writing + devrel). I write a lot of code and a lot of articles.
I think there is a lot of overlap in how people who write code or articles (documentation, books, etc.) use AI. On one side of the spectrum, you have people who just blindly input some prompt, accept the output and move on with their lives. You can likely predict how that is going for them (not great). On the other side of the spectrum, you have people who outright reject all AI and are continuing to plod on with how they have always done things.
In the center is a more reasonable approach that leverages AI to create without blindly accepting the output. This applies very much to writing.
The workflow that I've adopted over the past two years or so has been to leverage AI to help with the research and outline process. Once I'm happy with the structure I go and I write what I need to write.
This maps pretty closely to the code that I write. It's fine.
AI doesn't need to write better than the best humans. It only needs to be good enough to replace a large part of the mundane paid writing that used to support writers while they developed their craft.
People can tell when something they are experts in is being done poorly, but others can't, and it frustrates the experts. Then, those same people think something completely different is being done well despite what experts in that respective field say. It's like when someone from your family reads an article about your profession and then proceeds to tell you how your job works. lol
It's a really pervasive issue in society IMO. And a few prompts in someone's favorite LLM just reinforces it to people that don't know (any better|what they don't know).
[1] https://en.wiktionary.org/wiki/Gell-Mann_Amnesia_effect
Have you tested this systematically, or is it possible that you are experiencing survivorship bias? If there were any generative art pieces that you didn't notice, you would have thought that they were human-made. Therefore, all the pieces you identified were "obvious" to you. Not to mention false positives.
We can replace their statement by "I'm a visual artist and at generative art is often blindingly obvious.... To me....". In other words, they can notice some or most generative art but other people familiar with art can't, or at least can't as often as they do. I think what they are trying to say isn't too changed by that.
I have made digital art for 30 years and this all just sounds like what people use to say about digital art in general.
The main problem I see with generative art is not that you can tell it is generative. It is that most the art is shit. The same way if you gave a 1000 random people a blank canvas and paint, most the paintings would be shit too.
The counter example is there is a billboard that I see driving sometimes that is obviously AI generated graphics. It is so eye catching compared to any of the other billboards because most billboards are boring.
You are just puppeting the standard gate keeping bullshit to a new art form and personally I sick of reading this.
Who the fuck are you to say what art is or what art can be?
There may come players who focus on models that are good at writing for technical writing/docs , copyrighting ect but I think people will lean towards not using them and will rather have the "human touch" for the things that directly impact brand perception.
Keep in mind, every single AI company that is selling the idea that you don't need to hire designers and web design is "solved" have $100k retainer designers crafting their landing pages.
When I had Gemini 2.5 write a novel, it wasn't really objectively "good" by any stretch of the imagination, but while the prose was very purple and full of cliches and, well, bad writing I guess, it still felt ... subjectively good, at least for what it was.
Last week I did a run with GPT-5.6, and wow. On the one hand, it managed to produce 110,000 words that were "shockingly" coherent. The model was able to maintain state and plot lines and background details extremely well, much better than older models.
But I just don't like the prose. I haven't really liked _any_ prose that GPT-5.6 produces. It's significantly better at "instruction following" and keeping track of things, but, wow.
> “The sequence is consistent with their voluntary choices.” Mara enlarged the uncertainty field rather than the result. “It does not prove what happened to anyone we can’t observe. It does not prove contact did this. And it does not turn the Shard into treatment.”
GPT-5.6 in particular becomes so fixated on certain ideas like "consent" and epistemology, that by the end of the narrative, the prose and dialogue are all just "agent speech", despite the prompt/harness specifying that it's a _novel_ with narrative prose and such.
Interestingly, the model itself produces an accurate critique of its own output:
> The draft has become a *consent-centered medical, legal, and logistical procedural*. The important drift is therefore not that many events were omitted. It is that the retained events now prove a different thesis.
Which begs the question of if it would do better with a couple rounds of output -> critique -> revision. But I think I've had enough LLM prose for a bit...
It's possible to work around by generating short passages at a time with carefully constructed setup. But it's a real pain.
In this case, part of the experiment was to see what "oh-my-pi", a "fat and feature rich" LLM harness, could do when coupled with modern GPT, given a 6k~ word overview of a story, and told to come up with a plan to write/review/audit it, making use of subagents and all the fun new groovy LLMisms...
Part of the problem was just "it was basing its style off the last scene/chapter", but part of it was also that its instructions were constantly being "compressed" through repeated compactions. Even with the use of subagents, the "top level" agent's prompt was getting muddied, and in the "review" phase, it began to focus more and more on creating increasingly complex ledgers.
You can see this happen in the "plan" files it created for each chapter, looking at word count:
So it wasn't just that the prose was being based on an increasingly compressed "style" of the prior context window, but the planning for writing each scene was, itself, becoming fixated on the "continuity error correction" process itself, to the point where by the end, it had mostly forgotten about the prose part, and was completely fixated on ensuring maximum state continuity.This could definitely be fixed, but honestly, I've about had my fill of the "autonomous writing agent" goal. The idea was to make a model that could generate sufficiently interesting stories based on "vague premises" for my personal entertainment, but, "surprise", getting LLMs to actually produce both "new" and "coherent" content beyond what you specify is _hard_.
It seems like you really do need to just stay "in-the-loop" with every scene, and constantly provide correction/feedback, to correct the "semantic drift".
Or, gasp, I could just try writing things by hand again... :-)
I've used some Opus 4.5/4.6 via Antigravity and Sonnet by the web chat. I'm torn because as far as LLMs go, it does feel more ... "literate".
But maybe too literate, judging by how many people are complaining about "Claudeisms". I suspect Claude would be just as susceptible, if not more, to the sort of ... "moralizing" that GPT seems to gravitate towards (for lack of a better term).
Same with code, but that doesn't seem to matter anymore.
Code's ultimate goal, on the other hand, is to be run by computers.
For now. It also seems like we're still paying software engineers.
Are we at the point in the hype cycle to go "there's a skill for that" yet?
https://archive.org/details/robotsexsocialet0000unse
I remember, when I used ChatGPT the first time to write some E-mails and other documents, I suddenly realized that I sound exactly like corporate communications or most tech writers. That made me realize that these guys are also working off templates and basically produce variations of the same thing. Same as developers do.
In that respect, I'd be more than happy to read more direct quotes from Gemini and then read what a live hacker's take is on it. Slop provides the building blocks, and humans are the alchemists. Need both.
https://hn.algolia.com/?dateRange=pastMonth&page=0&prefix=tr... (not many returns for the last month, most are from me)
AI today is most effective when it's not vibing, but rather copiloting a skilled operator. I don't necessarily want my agent to build an entire system for me, even if it's ultimately the author of almost every line of code; I'm actively making decisions throughout. There's obviously a spectrum here but at most points on the spectrum the amount of assistance available is still a step change in the economics.
So too with writing.
The first rule of accelerating writing with AI is that you're not allowed to use a single word the AI suggests. Even if what the model comes up with is great, better than what you could have done, as soon as the AI suggests it it's poisoned. At least with current models, readers can detect LLM prose in the parts per trillion, and as soon as they do you've lost them.
The second rule of writing with AI is that AI encouragement is toxic. A structural consequence of RL is that models are exquisitely tuned to generate responses that make their users perceive value. We recognize this in a gross sense in "sycophancy", but the problem recurs fractally in at finer-grained levels, where stuff like "this part is really strong" will subtly allow the model to set a course for your writing and you'll confidently ship crap.
With those two rules in mind, models are incredibly valuable for writing, more valuable in my experience than the professional copywriters I've worked with. The trick is to get them to make suggestions at a higher level than just writing alternatives:
* Do the sentences in these paragraphs end with the new idea or information?
* Are the real actors in each sentence the grammatical subjects?
* From paragraph to paragraph is there a clear flow of topics, or are things jumping around?
* Is this piece crudded up with metadiscourse like "it's important to note"?
I've had a stack of notecards for ages that I took down from Joseph Williams "Style: Towards Clarity And Grace", the most programmer-brained writing book ever written, I love it very much. For the past year or so I've been feeding them through GPT and Claude one by one, and it's drastically increased the speed at which I can knock out a completed piece.
I think it's pretty hard to argue that AI isn't going to have an impact on the writing profession. It's just not the most obvious impact everyone assumes it will have, where it, like, writes whole op-eds or whatever. At least not yet.
In terms of what he's presenting here, I'll say this: his line of logic highlighting the "dual-mind problem" of writing seems, to me, like the most compelling aspect of this argument. When I write something (especially to a specific audience, including an individual), I must practice cognitive empathy if I want what I write to be consumed properly and effectively. This is a cold way of putting it, but even in text message responses which span a few words towards family or a friend, I can put quite a bit of thought into how it will be received, how they will read it, interpret it, etc. My cadence in just texting, alone, can shift dramatically from one message to a next based on who I'm sending it to, what I'm trying to convey, etc.
All of that is fine and not hard to understand, but I suppose the interesting part is this: when I'm just trying to get information across (i.e., instructions, directions, etc.), my messages will resemble something written by an AI. It's not enjoyable to consume, but that's not the point. But as soon as I want to add a bit of "fun" to a message, the task I'm performing is completely different. It's no longer just an exchange of information, but it's an attempt to invoke specific feelings, visualizations, memories, etc., in the other person/people.
I do have a hard time imagining what it will take for AIs to be able to do THAT effectively. I think they're fine at conveying information, but I think people are already becoming very aware that conveying information, in itself, is not enough to be effective. These LLMs don't "care" to "entertain" you with what they're writing at you about. They're just spitting out the mathematically derived facts with, seemingly, no meaningful ability to invoke deeper thoughts in their audiences' minds. And that might not seem particularly important outside the context of writing fiction, etc., but I think it's actually pretty critical even in "dry" settings, like explaining code, because the thoughts, feelings, emotions, etc., invoked through reading IS the output of reading (even if it doesn't seem that way).
Perhaps AI will be able to do this more effectively if they're trained on direct brain signals or something. Like, train the AI not just to convey accurate information, but also encourage it to do so in a way which stimulates different areas of the brain with different intensities based on the premise that, doing so, is actually what people are getting out of that text.
This doesn't ring true to me.
LLM prose can be greatly improved with the right prompts.
Now, prose with the right prompts may well still not be as good as a good human writer would write – but it is a lot better than what LLMs produce by default.
If the model can perform better with the right prompts, it suggest they haven't actually done everything they could in the post-training to maximise writing quality.
I watched a podcast with a cognitive scientist and one of main contributors to the theory of linguistic relativity, Lera Boroditsky.
She said something to the effect that, "in this very moment, we are speaking in ways that were never spoken before. We are saying things that no other person has said before...."
Language models are not sample efficient and cannot adapt to evolving language, unless it's documented in large amounts of examples.
So whatever isn't documented, whatever isn't in the training dataset or the rag corpus, the model will always be incredibly different in expression from humans.
I agree with the article. Even the frontier models call out that all the characters tend to sound the same. It also started at some point making huge changes to the core premise, and also adding characters willy nilly. The issues it called out with the plot (the ones it actually consulted me on) also made me realize how terrible of a writer I actually am.
Overall, it has been an interesting experience. I look forward to reading my own book!
I understand a lot better now why people are bemoaning KDP being filled with absolute garbage AI slop.
I did translation as a side job for about 20 years (helped me keep my language skills sharp, and some side income was welcome). I was good at it. But translation was one of the first professions to fall to AI (even before LLMs but especially since), and I saw increasingly that companies were willing to accept the clearly lower-quality work if it meant saving money. A couple of years ago I stopped doing any translation work altogether.
Sure, whoever is hired to translate Murakami's latest novel will be a human, a good writer first and foremost. But that translation work is a tiny fraction of the overall translation work contracted by companies.
The same thing is going to happen with writing. It won't disappear, but demand will drop by >80%.
Lol. If that were true software would've been a lot better historically... Model checkers don't scale to 90% of the software we write. Typically you have to 1) heavily abstract the program and 2) put it in some sort of harness to specify how you want to model the outside world (which will always fall short of practice). And probably 10 other workarounds since most model checkers are Research Grade Software™. Not saying they're not tremendously useful, but that bullet doesn't hold up at all.
It's sad, infuriating, discouraging. A deluge of slop drowning the last embers of authentic human creativity.
We get to learn the foibles and language of Claude and ChatGPT as a result. The slop is almost detectable if you provide no steering prompts about story structure, narrative structure, or stylistic cues. And most writers are not finetuning the weights to their LLMs explicitly.
If you invest time into doing all that (not really trivial stuff), the results will be better.
How long after that moment did the ChatGPT robots need to invent the time travel machine which let you come back to 2026 to confidently claim this?
For example the recent Skild AI fairly general imitation capability. It really is a huge amount of effort on increasing the generality of humanoid robots going on now and a lot of demonstrations coming out showing off progress.
With the current trajectory there is no reason to think it will stop soon. There is a lot of motivation to innovate and train and it is pushing things forward.
If you're not looking for evidence of improvement, or rather looking for failure cases, then you're not going to see that trajectory.
There's so much focus in the US on vocations but Americans have no idea how little mechanics, plumbers, etc. in other countries make. There are a variety of reasons for this but one of the biggest is that you can't charge $200/hour to fix a leak when $200 represents a double-digit percentage of a worker's monthly salary.
So the author may be correct, but for a different reason: unless writing starts being very valuable as a profession, it's unlikely the labs will spend significant resources making their AI models better at it.
Any given model will always have some distinct implicit voice that its biased towards for that infill content, and so a popular model will always become exhaustingly common, painfully familiar, and cliche. Users can use more elaborate prompts that shift the voice away from the most normative and towards some other nodes, but they need to put in special effort for that, and what people-at-scale specifically want from these tools is to put in very little effort, so we can expect that overwhelming number of casual and naive users will always be generating cliche slop with them.
Code escapes this problem not because of training but because it specifically benefits from cliche (boilerplate, patterns, etc) and so an model whose code "voice" reflects your own taste as a coder (or your toolchain's taste as a vibecoder) is going to feel like productive output rather than slop. But it's still cliche.
Even if LLM output has to largely follow some statistical rules, yet, first of all, some amount of randomness is normally injected during token generation, and, secondly same true for human speech.
> about what best fills in the gaps. It's not a training problem, it's an information theory problem, and it's not really surmountable.
This is not true, as LLM has internal knowledge storet in its weight. Unless you force it to produce 2000 words doc out of 3 word prompt, you would end up adding some sense information.
>Users can use more elaborate prompts that shift the voice away from the most normative and towards some other nodes, but they need to put in special effort for that, and what people-at-scale specifically want from these tools is to put in very little effort, so we can expect that overwhelming number of casual and naive users will always be generating cliche slop with them.
True, here I agree with you. But using finetuned or simply less popular models like Kimi, Hy etc. should take care of that.
Software can be checked for being 'written well' by compilers / linters etc. There is no equivalent for well-written natural prose. Spelling and grammar checkers haven't a clue about prose semantics.
It's slightly weird how confident writers are that it won't get improved.
There will always be people that work against their own profession and colleagues for short term gains
If you are taking steps to protect your brain maybe in a way you are also protecting your job?
Who knows how this all turns out years from now.
Given what seems like an increasingly inevitable deprecation of these outdated, lumbering nation-states, it seems to me that these two assertions are mutually exclusive.
One year ago, the writer Mark Lawrence posted an article [0] (featured on HN) describing how he generated 4 pieces of flash fiction with ChatGPT, asked 4 published authors (with a combined book sales of $15M) to also write a flash fiction piece, and then did a blind test asking respondents, for reach of the 8 pieces, (a) whether they thought it was written by a person or not and (b) an overall quality score. The result was that people (who were avid readers and skewed anti-AI) were no better than a coin toss at determining whether the piece was written by ChatGPT, and slightly preferred ChatGPT-written writing overall.
[0]: https://www.marklawrence.buzz/2025/08/the-ai-vs-authors-resu...
However, once it's understood that it wasn't written by a human it loses all value. I use this quote by Kane Parsons a lot but I think it summarizes why humans like art.
> If I see an element of the environment has been, you know they used generative fill or whatever to change something about the scene, even if--. It just shuts the part of my brain that wants to know more about that world and wants to look for details, 'cause I would assume if they're willing to make an arbitrary choice there, they can make an arbitrary choice with literally anything.
Specifically, people like dissecting art whether its writing, a movie or an image. It could be seeing the Chekov's Gun finally fire, and going back through the pages to see the author wrote it in there earlier and you missed it. It could be speculating about the unopened questions the author deliberately left there. This is how a lot of people at least in my circles engage with art and AI provides none of that. Even if the author specifically prompts that into it, those arbitrary choices are going to be made in certain spots that just disinterests me.
1: https://www.sciencedirect.com/science/article/pii/S096969892... 2: https://www.nature.com/articles/s41598-023-45202-3
Hmm...No. I actually enjoy generating short stories for myself - after light editing they are great; they are enjoyable irrespective of the provenance.
> I am appalled at the absolute shit [LLMs] spew as prose. They always follow the same robotic cadence and cliches, and sprinkle the same tired vocabulary all around.
This claim is demonstrably untrue, and central to the author's overall thesis.
Everybody treats this as something innevitable, like the movement of the tectonical plates.
Everybody knows this is being paid with the giant transfer of wealth from the working classes towards the asset owning class via profilgate government expense, ZIRP and QE policies from the FED and the covid so-called stymulus. This created, via Cantillon effect, inflation for consumers and a bizarre and absolutely abnormal deluge of capital for our financial system masters that let they play God and make us, the plebe, obsolete. And yet we don't question anything.
We are accepting as inevitable and uncontrollable something that only looks to be like that.
I know, we all have our lives, out of the confortable anonymate here, I am a enthusiastic AI Booster. Out of some true zealots, we are all hedging our bets to stay on the right side of the city walls of the new techno-feudalism. But, realistically, how many of us the Musks and Thiels will need after they obsolete most of us, no matter how much skills and knowledge anyone of us may have? Just betting to be on the side of the oppresors is a very long bet my friends.
Nothing about it is inevitable, nothing about it requires, as if it was a law of nature, to be unregulated and the people be damned.
Not so many centuries ago, we had peasants riotting and guillotining their opressors, we had proletarian revolutions. They didn't have encryption, the internet, cell phones, digital radio modes that allow you to talk to some other partisan group accross the globe using a cheap chinese SDR and a simple dipole hang between trees.
Things can be different. We still can force some democracy down their throats.
Yes write, but then orate.
Look at all of our AI feeds.
They want to be us so bad.
They absolutely cannot exhibit empathy. The definition of the word shows that:
> the action of understanding, being aware of, being sensitive to, and vicariously experiencing the feelings, thoughts, and experience of another
They don't experience feelings. They don't empathize.
It has no feelings, therefore no empathy.
Best come up out of the rabbit hole for some fresh air & sunshine brother.
AI will outdo people in all practical uses. We're already there for debugging and getting very close for coding, and we're in the middle of the largest investment in human history to expand that to everything else.
If we do a good job of alignment, AI will treat people like those cats "in charge" of train stations in Japan: our every need will be accommodated, but we won't be controlling things we don't understand.
What makes you think that there are economic incentives to build an AI that has the same capabilities as the human brain?
The current scaling story is to intentionally take a system that objectively doesn't and sell it as if it does.
I'm not cocky enough to bet against it. Especially since now AI is solving frontier math problems, debugging better than humans, and writing most of the content posted to Hacker News.