I think that this requires some nuance. Was the post generated with a simple short prompt that contributed little? Sure, it's probably slop.
But if the post was generated through a long process of back-and-forth with the model, where significant modifications/additions were made by a human? I don't think there's anything wrong with that.
I remember this was back in 2023, when ChatGPT had first launched, and I had a manager whose English was not very good. He started sending emails that felt like they were written by a copywriter. And the messaging was so hard to parse through because there's so much ChatGPT fluff around it. Very quickly we realized that what he was saying was usually in the middle somewhere, but we'd have to read through the intro and the ending of the emails just so that we couldn't miss anything. It felt like wasting 2-3 extra minutes per team member.
I really like Oxide's take on AI for prose: https://rfd.shared.oxide.computer/rfd/0576 and how it breaks the "social contract" where usually it takes more effort to write than to read, and so you have a sense that it's worth it to read.
So I get the frustration that "ai;dr" captures. On the other hand, I've also seen human writing incorrectly labeled AI. I wrote (using AI!) https://seeitwritten.com as a bit of an experiment on that front. It basically is a little keylogger that records your composition of the comment, so someone can replay it and see that it was written by a human (or a very sophisticated agent!). I've found it to be a little unsettling, though, having your rewrites and false starts available for all to see, so I'm not sure if I like it.
I don't like AI writing because it's bad writing. It's convoluted and inefficient and doesn't get to the point. If someone writes something that feels like AI, it doesn't matter if it was or not because it's still bad writing. I'm not talking about having a cliche here and there, but rather when the text is just incredibly inefficient.
> AI-generated code feels like progress and efficiency, while AI-generated articles and posts feel low-effort
I've noticed that attitude a lot. Everyone thinks their use of AI is perfectly justified while the others are generating slops. In gamedev it's especially prominent - artists think generating code is perfectly ok but get acute stress response when someone suggests generating art assets.
There's real reasons behind the huge divide in how the median programmer views LLMs vs how the median artist views image models.
1. Most programmers are better than SOTA LLMs, while most artists can't match the rendering quality of a SOTA image model. Artists rightfully see image models as a way bigger threat than we see LLMs.
2. While it's true that LLMs are trained on unlicensed code and image models are trained on unlicensed art, a lot of the publicly available code was essentially released with a license of "here's some code, you can use it for whatever"
3. Code is seen as a means to an end, while art is an end in itself. Few people, even among professional programmers love programming, or program recreationally (this forum obviously has disproportionately many). Most artists love the process of making art.
> For me, writing is the most direct window into how someone thinks, perceives, and groks the world. Once you outsource that to an LLM, I'm not sure what we're even doing here. Why should I bother to read something someone else couldn't be bothered to write?
Because writing is a dirty, scratched window with liquid between the frames and an LLM can be the microfiber cloth and degreaser that makes it just a bit clearer.
Outsourcing thinking is bad. Using an LLM to assist in communicating thought is or at least can be good.
The real problem I think the author has here is that it can be difficult to tell the difference and therefore difficult to judge if it id worth your time. However, I think author/publisher reputation is a far better signal than looking for AI tells.
I roll my eyes every time I see a coworker post a very a long message full of emojis, obviously generated by a LLM with 0 post editing. Even worse when it's for social communication such as welcoming a new member in the team. It just feels so fake and disingenuous, I might even say gross.
I don't understand how they can think it's a good idea, I instantly classify them as lazy and unauthentic. I'd rather get texts full of mistakes coming straight out of their head than this slop.
It's pretty much over for the human-internet. Search was gamed, its usefulness has plummeted, so humans will increasingly ask their LLM of choice and that LLM will have been trained on the content of the internet.
So when someone wants to know something about the topic that my website is focused on, chances are it will not be the material from the website they see directly, but a summary of what the LLM learned from my website.
Ergo, if I want to get my message across I have to write for the LLM. It's the only reader that really matters and it is going to have its stylistic preferences (I suspect bland, corporate, factual, authoritative, avoiding controversy but this will be the new SEO).
> "It's pretty much over for the human-internet ... chances are it will not be the material from the website they see directly..."
Well, we're here. HN manages to call me back to the community nearly every day with articles I would not have otherwise found by search or AI.
No doubt web search just feels inadequate by the qualities of chat initiated search. But the results of chat can lead to websites and books and all manner of media consumed by the ChatAi. The pathways are still there to find those pages.
What I think you're talking about are business statistics in consumer industry. Maybe you're thinking of a person who drinks Bud Lite and eats Lays potato chips (highly ranked sales) and envision they'll use ChatAi apps and never use a search engine again. If that's the demographic you want, then writing a website is prolly not going to reach them anyway. If you want to reach that audience, then you need to become part of that mass consumer media ecosystem. You need to create/invent virality--it's the one shared element consumer society revolves upon and industry can't get enough of.
I laugh every time somebody qualifies their anti-AI comments with "Actually I really like AI, I use it for everything else". The problem is bad, but the cause of the problem (and especially paying for the cause of the problem)? That's good!
I agree with the general statement, if you didn’t spend time on writing it, I am not going to spend time reading it. That includes situations where the writer decides to strip all personality by letting AI format the end product. There’s irony in not wanting to read AI content, but still using it for code and especially documentation though, where the same principle should apply.
>I can't imaging writing code by myself again, specially documentation, tests and most scaffolding.
Doesn't ai;dr kind of contradict ai generated documentation? If I want to know what claude thinks about your code I can just ask it. Imo documentation is the least amenable thing to ai. As the article itself says, I want to read some intention and see how you shape whatever you're documenting.
(AI adding tests seems like a good use, not sure what's meant by scaffolding)
Many engineers suck at writing. I'm fine with AI prose if it's more organized and information-dense than human prose. I'm sick of reading 6 page eng blogs to find a paragraph's worth of information.
Yeah, I use LLM agents extensively for coding, but I have never once allowed an LLM to write anything for me. In the past month, I literally wrote 40,000 words of researched essays on various topics, and every single word was manually written, and every source manually read, myself. Writing is how I think, how I process information, and it's also an activity where efficiency is really not the goal.
Thoughts with the people who use AI to help construct their thoughts because their cognitive decline impacts the ability to construct words and sentences, but still enjoying the production of content, blogging and th indieweb.
These blanket binary takes are tiresome. There is nuance and rough edges.
I use a technique where LLMs help me write, but the final output is manual and entirely mine. It's a bit of heavy process, but I think it blends the power of LLM and authenticity of my thoughts fairly well, I'll paste in my blog post below (which wasn't produced using this method, hence the rambly nature of it):
If you care about your voice, don't let LLMs write your words. But that doesn't mean you can't use AI to think, critique and draft lots of words for you. It depends on what purpose you're writing it for. If you're writing an impersonal document, like a design document, briefing, etc then who cares. In some cases you already have to write them in a voice that is not your own. Go ahead and write these in AI. But if you're trying to say something more personal then the words should be your own, AI will always try to 'smooth' out your voice, and if you care about it, you gotta write it yourself.
Now, how do you use AI effectively and still retain your voice? Here's one technique that works well: start with a voice memo, just record yourself maybe during a walk, and talk about a subject you want, free form, skip around jump sentences, just get it all out of your brain. Then open up a chat, add the recording or transcript, clearly state your intent in one sentence and ask the AI to consider your thoughts, your intent and ask clarifying questions. Like, what does the AI not understand about how your thoughts support the clearly stated intent of what you want to say? That'll produce a first draft, which will be bad. Then tell the AI all the things that don't make sense to you, that you don't like, just comment on the whole doc, get a second draft. Ask the AI if it has more questions for you, you can use live chat to make this conversation go smoother as well, when the AI is asking you questions, you can talk freely by voice. Repeat this one or two more times, and a much finer draft will take shape that is closer to what you want to say. During this drafting state, the AI will always try to smooth or average out your ideas, so it is important to keep pointing out all the ways in which it is wrong.
This process will help you with all the thinking involved being more up-front. Once you're read and critiqued several drafts, all your ideas will be much more clear and sort of 'cached' and ready to be used in your head. Then, sit down and write your own words from scratch, they will come much easier after all your thoughts have been exercised during the drafting process.
This is the root cause of the problem. Labeling all things as just "content". Content entering the lexicon is a mind shift in people. People are not looking for information, or art, just content. If all you want is content then AI is acceptable. If you want art then it becomes less good.
It became "content" when it became a vehicle to serve ads for companies and a means to promote yourself for individuals. So, yes, "content" isn't really made for other people. It's made to get something out of other people - mostly their money.
When people put together memos or decks in the last, even if that weren’t read very carefully, at least they reassured management that someone had actually things through. But that is no longer a reliable signal.
> I can't imaging writing code by myself again, specially documentation, tests and most scaffolding
> Why should I bother to read something someone else couldn't be bothered to write?
Interesting mix of sentiments. Is this code you're generating primarily as part of a solo operation? If not, how do coworkers/code reviewers feel about it?
81 comments
[ 3.9 ms ] story [ 60.2 ms ] threadBut if the post was generated through a long process of back-and-forth with the model, where significant modifications/additions were made by a human? I don't think there's anything wrong with that.
How we can tell that this wasn't written by an LLM.
So I get the frustration that "ai;dr" captures. On the other hand, I've also seen human writing incorrectly labeled AI. I wrote (using AI!) https://seeitwritten.com as a bit of an experiment on that front. It basically is a little keylogger that records your composition of the comment, so someone can replay it and see that it was written by a human (or a very sophisticated agent!). I've found it to be a little unsettling, though, having your rewrites and false starts available for all to see, so I'm not sure if I like it.
I've noticed that attitude a lot. Everyone thinks their use of AI is perfectly justified while the others are generating slops. In gamedev it's especially prominent - artists think generating code is perfectly ok but get acute stress response when someone suggests generating art assets.
1. Most programmers are better than SOTA LLMs, while most artists can't match the rendering quality of a SOTA image model. Artists rightfully see image models as a way bigger threat than we see LLMs.
2. While it's true that LLMs are trained on unlicensed code and image models are trained on unlicensed art, a lot of the publicly available code was essentially released with a license of "here's some code, you can use it for whatever"
3. Code is seen as a means to an end, while art is an end in itself. Few people, even among professional programmers love programming, or program recreationally (this forum obviously has disproportionately many). Most artists love the process of making art.
Shouldn’t we bother to write these things?
Because writing is a dirty, scratched window with liquid between the frames and an LLM can be the microfiber cloth and degreaser that makes it just a bit clearer.
Outsourcing thinking is bad. Using an LLM to assist in communicating thought is or at least can be good.
The real problem I think the author has here is that it can be difficult to tell the difference and therefore difficult to judge if it id worth your time. However, I think author/publisher reputation is a far better signal than looking for AI tells.
I don't understand how they can think it's a good idea, I instantly classify them as lazy and unauthentic. I'd rather get texts full of mistakes coming straight out of their head than this slop.
So when someone wants to know something about the topic that my website is focused on, chances are it will not be the material from the website they see directly, but a summary of what the LLM learned from my website.
Ergo, if I want to get my message across I have to write for the LLM. It's the only reader that really matters and it is going to have its stylistic preferences (I suspect bland, corporate, factual, authoritative, avoiding controversy but this will be the new SEO).
We meatbags are not the audience.
Well, we're here. HN manages to call me back to the community nearly every day with articles I would not have otherwise found by search or AI.
No doubt web search just feels inadequate by the qualities of chat initiated search. But the results of chat can lead to websites and books and all manner of media consumed by the ChatAi. The pathways are still there to find those pages.
What I think you're talking about are business statistics in consumer industry. Maybe you're thinking of a person who drinks Bud Lite and eats Lays potato chips (highly ranked sales) and envision they'll use ChatAi apps and never use a search engine again. If that's the demographic you want, then writing a website is prolly not going to reach them anyway. If you want to reach that audience, then you need to become part of that mass consumer media ecosystem. You need to create/invent virality--it's the one shared element consumer society revolves upon and industry can't get enough of.
https://noonker.github.io/posts/2024-07-25-i-respect-our-sha...
I can take the other person's prompt and run it through an LLM myself and proceed from there.
Doesn't ai;dr kind of contradict ai generated documentation? If I want to know what claude thinks about your code I can just ask it. Imo documentation is the least amenable thing to ai. As the article itself says, I want to read some intention and see how you shape whatever you're documenting.
(AI adding tests seems like a good use, not sure what's meant by scaffolding)
These blanket binary takes are tiresome. There is nuance and rough edges.
If you care about your voice, don't let LLMs write your words. But that doesn't mean you can't use AI to think, critique and draft lots of words for you. It depends on what purpose you're writing it for. If you're writing an impersonal document, like a design document, briefing, etc then who cares. In some cases you already have to write them in a voice that is not your own. Go ahead and write these in AI. But if you're trying to say something more personal then the words should be your own, AI will always try to 'smooth' out your voice, and if you care about it, you gotta write it yourself.
Now, how do you use AI effectively and still retain your voice? Here's one technique that works well: start with a voice memo, just record yourself maybe during a walk, and talk about a subject you want, free form, skip around jump sentences, just get it all out of your brain. Then open up a chat, add the recording or transcript, clearly state your intent in one sentence and ask the AI to consider your thoughts, your intent and ask clarifying questions. Like, what does the AI not understand about how your thoughts support the clearly stated intent of what you want to say? That'll produce a first draft, which will be bad. Then tell the AI all the things that don't make sense to you, that you don't like, just comment on the whole doc, get a second draft. Ask the AI if it has more questions for you, you can use live chat to make this conversation go smoother as well, when the AI is asking you questions, you can talk freely by voice. Repeat this one or two more times, and a much finer draft will take shape that is closer to what you want to say. During this drafting state, the AI will always try to smooth or average out your ideas, so it is important to keep pointing out all the ways in which it is wrong.
This process will help you with all the thinking involved being more up-front. Once you're read and critiqued several drafts, all your ideas will be much more clear and sort of 'cached' and ready to be used in your head. Then, sit down and write your own words from scratch, they will come much easier after all your thoughts have been exercised during the drafting process.
This is the root cause of the problem. Labeling all things as just "content". Content entering the lexicon is a mind shift in people. People are not looking for information, or art, just content. If all you want is content then AI is acceptable. If you want art then it becomes less good.
> Why should I bother to read something someone else couldn't be bothered to write?
Interesting mix of sentiments. Is this code you're generating primarily as part of a solo operation? If not, how do coworkers/code reviewers feel about it?