For the record, I never use AI in writing. First, it's no fun, and that's really reason enough. It'd be like sending a bot to watch a movie for me. Second, I'm better than it is, and it would just be an exercise in frustration micromanaging its every word. I might as well just type what I want it to say.
And third, most importantly, who would read anything I wrote if they could just generate it themselves? The goal is to be better.
This should be embarrassingly obvious to everyone, but allegedly there are hoards of vibecoders and other slop producers to whom it never occurs (if it's not astroturfers pretending to be that). If prompt slinging is your only expertise, you already are redundant.
I still have to make a bunch of decisions based on its research, so it still turns out to be hours of work that can go into it.
Granted, I can probably optimize some things by encoding my style preferences. But so far it has been a lot more involved than „make me an app, make no mistakes“.
That‘s why I‘d call it in a large part my work. It‘s my vision, in the same way an architect that has never personally laid a brick can call a building „their work“.
But the question is how often are you better at all the tasks where you don’t have maximal time, attention, and energy - and it would make no sense to invest it.
Let the reader decide the value of what they read at face value.
I honestly feel this ai disclosure thing is just pure mindless elitism and worse - entitlement - from readers. "Crappy" writing has always existed, deal with it. Stop reading as soon as you want
Take saltwatwer and cell. After some threshold of saltiness is passed, the water flow reverses from into the cell to away from the cell. And the cell frikkin dies.
You never bothered explaining why saltwater is ai writing and readers are cells, nor what the parallel of immersion or osmotic pressure would be.
You lazily wanted me to fill in the details and be convinced. Why did you create this bad argument and leave it unlabelled.
Web content discovery is almost universally funded by advertisers… while yes the core technology is requests based, the things you get recommended are harder to control.
Unfortunately slop is very cheap to produce compared to human generated content and literally endless, so content providers have an incentive to push as much as their users will tolerate.
The problem is that AI texts are flooding the web and hijacking the attention of humans, their most precious and scarce resource
> Stop reading as soon as you want
Then you already lost time
I don't care whether my favorite author sweated for months facing a typewriter, of he effortlessly dictated the final form of the book in one sitting to a secretary while sipping mojitos.
I think my issue with AI has more to do with the signal it sends: reading takes effort, particularly literature, and I use the author's name as a proxy to judge whether to invest that effort myself. Nothing bad in selling dollar store crap, but it's bad to put 'Nike' on it.
Your individuality is what you sell as an author. I can get access to the LLM without you.
my reading is: if you use AI to help you write, then i can't know how much of the work is yours and how much is AI. therefore, when AI helps i have to expect the worst and assume that it is mainly AI and your input is very little. consequently, don't use AI at all, or the work is no longer yours.
i think that's a pretty good argument. it's not about the effort you made but about the amount of control you have over the text. and, as you say, the signal it sends. so i think you agree more with the article than you say.
what is weird is the title: don't say you use AI for writing, but then in the text it says: don't lie. if you can't do either then you can't use AI for writing, so why not just say it directly: don't use AI to help you writing.
>was I tempted to use AI to speed things up? No. It would be like hooking a motor to a stationary bike and calling that exercise. It would be like taking a helicopter to the top of Everest and saying that I summited.
And you can read a car maintenance book. That doesn't mean you can fix your car now.
The author vouched the LLM output using their experience, that's what you get. Unless you are as experienced in their domain, it will take you time to figure out if the output of your LLM prompt is correct or non-sense.
It doesn't work for me sorry, because you wouldn't accept a book by John, a friend of Hemingway, as a Hemingway book no matter how much he assisted in editing. Nor a Picasso museum exhibition is by Marie because Marie chose which paintings to display.
Authorship and edition are different claims.
The creator still gets the credit.
LLMs can just be the part that accelerates laying the code down.
I think folks are just too emotional over a tool that we are ignoring drawing similarities on purpose. That or just different audiences, hackers vs professionals. The latter just meaning being payed and usually working in a team. The styles can be different and the value placed on crafted code vs results.
Actually, they all get credit, at least in Hollywood
and how do you handle the reverse? use AI as the editor for a hemmingway book. is it still hemmingway?
I mostly agree with that, with some caveats (in short, there's an uncomfortably thin line to appropriating the curated work, consciously or not).
>and how do you handle the reverse? use AI as the editor for a hemmingway book. is it still hemmingway?
I'd be comfortable saying at least that it's less misleading to declare that book as written by Hemingway. Doing so is more in line with the social expectations that come with having the author's signature on it.
Traditional authorship is self-vouching: writing a coherent book on a complicated subject is hard. You can't exactly bullshit your way through 500 pages of car maintenance minutia. If a book actually manages to make its way all the way to a book store, there is a pretty good chance it is worth reading.
LLMs change this equation. Any idiot can prompt an LLM into writing reasonable-sounding slop, so any idiot can now write a book on any subject. Combine that with self-publishing and print-on-demand, and suddenly all bets are off.
If 99% of LLM-written (or LLM-"assisted") books on a subject are garbage you aren't going to buy a book and hope it is the 1% - it makes far more sense to save yourself the money and get the mediocre answer by prompting the LLM yourself. Want expert information? Just buy an LLM-free book written by an actual expert.
It’s not like people feel the need to explain that they used a spellchecker or thesaurus or googled the correct use of an idiom, etc.
There seems to be a general need for some people to dunk on valuable AI use by refusing to acknowledge that there are a many ways to use a tool. (Similar story on using AI for coding.)
Echoing a comment from above, why would I care whether a sentence was formulated by AI or John or a ghostwriter or John who asked Jane for feedback before rewriting? I care about the content. If I don’t like the way it’s written or if I’m irked by how it is written (ooh, an emdash — how embarrassing!) then I don’t need to read it.
Personally, I’m much more annoyed when I click on an article that sounds interesting, and the author tries to show off their penmanship by starting with five pages of “a history of X” or a tangential but redundant personal story before getting to the point. IMHO most non-fiction writing on the web could be accomplished in bullets. But again, that’s just my very personal preference…
My line is clear, if you use copy paste the AI output that's not your writing. I am okay with AI collaboration - it detects the errors, you decide what to do with them.
I want to read code by Abrash, Peyton Jones and Karpathy, not Claude's output based on a prompt from a third rater.
If you send me AI generated writing, I will have my AI agent read it and respond to it.
Meanwhile I will use my limited human time to engage with humans and human created content.
So, I agree with the moral stance of the article, and I personally take all the Ls from all the CV rejection, and the lower grades because I write the paper myself. Just realize that you are choosing life on hard mode for negative societal and "dominance hierarchy" benefit.
If the author of this write up actually used AI for writing he would have way more than just word. Because you can definitely tell AI output that somebody put no effort into from output that somebody put in a ton of effort into.
It can be as much of a difference as between artistic photograph and a photo from a photo-trap installed in the forest or from a speed camera.
"One number. Four completely different stories. The number is engineered to include all of them, because including all of them is what produces the 49 percent."
I decided to fact-check a statement ("CNN’s May 2026 survey found the share of Americans spontaneously naming gas prices as their top economic problem rose from 5 percent to 23 percent in a single year, with food costs cited almost as often") and it was incorrect (food costs were in fact cited more often than gas prices). Since the first thing I checked was wrong, I decided it wasn't worth my time reading the rest of the article. It was, as they say these days, slop.
It felt like a little bit of my time had been stolen. If a disclosure had been at the top, it would have been more of a caveat emptor situation.
It is very common to see that any interesting thought gets immediately tagged like AI slop and the real AI slop wins. Try an A/B test and you shall see that AI actually wins because of the people who hate AI. Most people cannot distinguish between a human and a AI written post and yet those same people want to be judgemental. And the people who are against AI and say "its just the next token generator and I don't use it" and yet use autocomplete on their mobiles are just duplicit. And yes AI is the next-token-generator, we have no proof that most humans were not brainwashed to become the same.
The “humans are just as bad” argument holds no water. Humans have many infirmities, but you can form relationships of trust with people who earn trust and deserve it. You can hold people accountable, personally and publicly. With the exception of rare criminals and the severely mentally ill, you understand people and you can work with them.
Anyway, if you hire someone to pretend to be you, that’s fraud. That’s being a scammer. Don’t do that. For the same reason, don’t let AI write in your name.
Or if you do, remember that I warned you how your reputation would collapse and people would stop taking your work seriously.
> If AI deeply collaborates with you to write something, why am I saying you shouldn’t say you used AI? Because all I have is your word for it that you did any work at all.
So using author's logic, I should not trust them when they say they never use AI for writing, because all we have is their word.
> “I’m a skilled liar. I frequently tell lies. But don’t worry, I wouldn’t lie to you!”
Interesting how saying you used any amount of AI instantly labels you as "a skilled liar" to the author.
Their arguments are mostly addressed by proper, clear attribution. "My sister helped with my homework essay" deserves distrust withour further clarification.
Comparing LLM usage to lying is a fun perspective, but most of the lying happens in attribution. Their moral against lying also seems silly.
But I'm not ashamed to say that I used it last week in a chat conversation with a recruiter to turn this:
1. I just said I'm hard of hearing and prefer text.
2. If it's only two minutes you can darn well send email.
As I mentioned, I'm hard of hearing and phone calls are difficult for me —
I find I miss things and it's frustrating for both sides. If it's just a
couple of minutes' worth of information, an email works great and I can
give you a thoughtful response. Happy to go from there!
If it makes James Bach think I'm a liar, that's a price I'm willing to pay.
>This recruiter didn't deserve my authentic voice or my personal toil
Yet they deserve inauthentic politeness from a chatbot? It just sounds like "fuck you" with more steps.
And it wasn't inauthentic. I wanted to be polite and inoffensive, but I didn't know how.
Maybe it’s not your goal to be seriously? But if it is…
If one uses the first person in one’s writing, it follows that the words are their own.
Anything else is disingenuous.
more at 11
Because in his view, if you use AI and don't disclose it, you're a liar. And if you use AI and disclose it, he won't trust you anyway.
If you’ve “written” something with AI, I have idea if you even read it, thus I have no idea if it even really reflects your thoughts. And I don’t care what a computer has to say, I care what a human has to say.
The problem we have now is determining if the person actually wrote it. It suddenly got a lot easier for people to get someone else to generate text. And there are a lot more lazy humans than skilled writers.
At a more fundamental level, if AI generated it then I have no trust it is actually true or reflects facts or matches reality. It's insulting to throw AI slop at us because you expect us to read something you didn't bothered to write or perhaps even read. The text is probably all wrong with a veneer of well sounding verbiage, and potentially is created to drive engagement instead of actually communicating useful information.
I've put that sentence in the title above because (per https://news.ycombinator.com/newsguidelines.html) it seems less misleading.
My post was much more than saying “don’t use AI.”
I use AI. I even use AI adjacent to my writing. But with me, you always know that my written words are directly from me. They sound like me. I pretty much write how I talk, albeit more grammatically.
My post was about why you are taking a big risk with your reputation/brand if you let AI draft your written communication on your behalf.
First, it's not a good look to start off your rebuttal with an ad hominem. Based on the high standards you speak to in your post, and your concern about reputations, I would have expected more.
> I use AI. I even use AI adjacent to my writing. But with me, you always know that my written words are directly from me. They sound like me. I pretty much write how I talk, albeit more grammatically.
To play the role of cynic, what does "AI adjacent to my writing" even mean? You wrote in your post:
> If AI deeply collaborates with you to write something, why am I saying you shouldn’t say you used AI? Because all I have is your word for it that you did any work at all.
Who decides if your "adjacent" use of AI was deep or meaningful? You wrote:
> Unless I know exactly how you did that work– which would have to have been from personal observation– I can’t know the truth.
You're admitting that you use AI but only "adjacent to [your] writing". Even if you explain what that means, how can I trust you? I wasn't there to personally observe how you used AI.
If you consult with AI in any fashion to flesh out an idea, test your arguments, etc., someone with an even more extreme position could say, "Well, even if he wrote the words, AI must have influenced his writing in some way, perhaps even unconsciously, and therefore everything he wrote is tainted."
You misunderstood my blog post in an odd way that's hard for me to account for unless you just didn't read carefully. Hence, my annoyed quip.
I wrote in my policy, referred to in the post, that I never have AI draft anything for me. But I was quite explicit that I may use AI to aid my research in various ways-- just as I use Google search and other tools. I might create a monte carlo simulation to test my understanding of probability before making a bold pronouncement about statistical matter, too.
So, what I mean by adjacent is "next to" but not "on top of" or "inside." That's what most people mean by adjacent. I don't have AI draft any text for me. I type it all myself. I choose the words and sentence forms and rhetorical structure and overall strategy myself. That's how you know it's me and not some AI masquerading as me.
Let's walk through this very carefully.
Your post states:
> If AI deeply collaborates with you to write something, why am I saying you shouldn’t say you used AI? Because all I have is your word for it that you did any work at all. Unless I know exactly how you did that work– which would have to have been from personal observation– I can’t know the truth.
Now let's look at your policy, which states:
> I allow myself to use AI to help develop or critique ideas or to critique text. I hate to be wrong. If AI can help me be less wrong, I welcome it. Although I always start with my own ideas, I might ask AI to challenge those ideas, or independently research the topic. I would then look over its work and decide if I want to cover a new topic or angle that the AI may have indicated.
> I might also ask AI to review the final text and spot typos or sloppy writing. In other words, I can use AI the same way I would let a human colleague help me write a piece for which I would nevertheless declare myself sole author.
Do you not see the problem here? You admit that you use AI as a stand-in for a human editor. You admit that you use AI to "help develop or critique ideas or to critique text". You admit that you use AI to do research. You even admit that this process might lead you to decide "to cover a new topic or angle that the AI may have indicated."
Do you not believe that some of this falls afoul of your own standard ("if AI deeply collaborates with you to write something...")? If you don't, please explain why you believe AI helping you develop and critique your ideas and text is not "deep" collaboration. And please explain how anyone reading your work, without the "personal observation" you referred to in your post, would know how much influence your use of AI had on the words you wrote.
This is such a lazy and tired way of talking around someone. You started off with a rude, dismissive statement which was intended to cause offense. Now that offense was taken and returned, you’re playing the “you aren’t being polite enough” game.
It’s transparent, and you are not entitled to politeness.
People are allowed to set their expectations/standards but in 2026 taking the position that use of AI is lying (when not disclosed) and trust destroying (when disclosed) is basically going to set you up for a lot of disappointment. It's just unrealistic.
For better or worse, AI is being used everywhere and it's harder and harder to spot, especially when the use is "thoughtful". Your only real defense is to think critically about the content you're consuming to determine whether it's accurate and has value.
It's not, only suckers think otherwise. The more you consume, the easier it gets to spot them, and also you get bored of it quicker. Which is fine for an actual tool.
IMO: if you think it's problematic if people could spot AI tool marks, you're not actually viewing or using AI as a tool. Rocket scientists aren't ashamed of using high end 5-axis CNC or SLS laser metal 3D printers to make rocket engines. Good machinists can tell how they were made, and that should mostly inspire confidence. If someone thinks the tool marks for a specific type of a tool needs to be hidden for the artifacts to be trusted, there has to be something wrong somewhere with the tool or how it's used, or both. Likely both.
Anyone looking at a Boeing 787 can tell that it flies on a pair of turbofan engines, and it's cool. Most people looking at AI images can tell it's generated using AI, and some can even identify models used, and that is NOT cool. That should be a strong enough sign that something is wrong with AI.
This was on the front page for a while yesterday. A decent amount of discussion. I'm quite sure this was produced using AI.
I was the only person who mentioned it.
AI is really easy to spot if it's being used to do all the work. It's less easy to spot when it's being used as a starting point, for editorial passes, concept development, argument refinement, etc.
Better to be Socrates dissatisfied than a fool satisfied.
It is absolutely possible to produce an insightful article using AI. But it intakes skill and dedication few people have.
But AI is only the latest continuation of what you're describing. The internet has been full of slop (clickbait, SEO-bait, etc.) and propaganda/disinformation for many years before AI was even a thing. Social media gave every person on the planet with a heartbeat and internet connection a publishing platform over a decade ago.
The only realistic approach for dealing with this is to exercise critical thought when you consume content. And if the massive volume of content we're flooded with is problematic, narrow the sources from which you consume content and consume less of it. Get off social media. Disavow YouTube. Don't doom scroll the news. And so on.
The likelihood that the author has consumed and trusted content that was produced using AI in some form, and not even realized it, is close to 100%. It's literally everywhere these days, and not everyone using it is using it to do all the work. But it leaves little hints that it was involved.
There are frequently posts that hit the front page of HN that have numerous AI fingerprints that produce discussion devoid of any comments questioning whether they were produced using AI. And HNers are probably one of the groups more likely than the general population to be able to identify AI content.
Isn't that, ironically, exactly the sentiment that motivates people to use AI to produce content?
I think your comment hints at the reality: we're all just increasingly lazy. We want to minimize how much time and effort it takes to produce content, and at the same time we want to minimize how much time and effort it takes to engage with it.
It's a vicious cycle.
This is the part the original human poster is assumed to have screened as a first step, not the audience, particularly if the audience is unfamiliar with the subject (such as a guide, etc).
I literally came across a guide online from a user who wasn't a spammer, who disclaimed they haven't even read the very guide they posted as an article on their website, as it was LLM generated. At least that user put up a disclaimer but why would I trust such a guide, given my and others' extremely inconsistent experience with the veracity of LLM output and as someone coming to the guide to learn (ie: not a domain expert)? Overwhelmingly other users don't put up such disclaimers so we don't even get to know whether they've vetted anything.
Trust is the key thing. To continually erode reader trust means you're putting the burden at every step on the reader. Sure, one should always apply critical thinking to even human output but there is an implicit, baseline assumption that with human output they're at least familiar with what they've output (whether they're lying or telling the truth or ignorant but honest). LLMs meanwhile handle ground truths in a flaky way, such as when they'll hallucinate quotes from even articles they claim to have read and cited. And the most common models users are using are the cheapest/free ones anyway, only compounding the accuracy issues.
Imagine you went to a library assuming authors, publishers and library staff have done some minimum due diligence only to find the library is being replaced rapidly with books that no one in the chain has read.
No one can be a domain expert in every single thing they encounter, which is why we place trust in others to varying degrees to fill in the gaps based on their experience and knowledge, even if you're a dyed in the wool skeptic. When increasingly what we encounter isn't being vetted as a basic first step then it's a waste of time and rude to the audience, which only decreases peoples' tolerance for bullshit and increases cynicism (which we could use less of).