[manphone]

It’s the same reason why everyone doesn’t wanna read LLM generated blog posts. The agreement used to be generally that you would spend more time writing than I would have to spend reading and when the agreement changes the quality changes as well.

[27183]

Writing is thinking. If you don't spend time writing, then you didn't spend time thinking. LLMs don't think. Consequently, if you outsourced your writing to an LLM the resulting artifact was born of a thoughtless process. Instead of engaging your readers in your thought process you're tricking them with a puzzle. Readers will try to understand what you're thinking about (at least the non-passive ones will). This activity is a dead end for LLM output.

[jgraham]

It's not that writing _is_ thinking, but (effective) writing _requires_ thinking. It forces you to take a collection of lose and incoherent thoughts and reify them into an argument that can hold its own weight.

[sph]

The art of essay writing allows an author to ascertain how well they understand a topic and work out the flaws just by trying to put it into words. Paul Graham speaks about this.

[Leonard_of_Q]

Is writing thinking? Good writing requires thinking but does that also go for sloppy writing? What is thinking? This is the subject of a discussion [1, 2] I'm in - which in true HN style has been greyed-out because of ${reasons} - regarding the question whether these models 'think'. Your comment on writing [being] thinking supports the idea that what these models do is a form of thinking: selecting the next most likely descriptor given the current problem space including all previously added descriptors.

[1] https://news.ycombinator.com/item?id=48884193

[2] https://news.ycombinator.com/item?id=48890755

[efficax]

writing is a form of thinking, but there are others

[gavinsyancey]

Yeah -- if it wasn't worth your time to write the post, why do I think it should be worth my time to read it?

[duggan]

Fabulous use of meta-irony, bravo.

[UltraSane]

Reading the output of a LLM prompt someone else wrote isn't useful as I can run the prompt myself. Having the prompt is actually more useful because I can run it in any LLM I want.

[pixl97]

Look at Mr infinite tokens over here.

[locknitpicker]

> It’s the same reason why everyone doesn’t wanna read LLM generated blog posts. The agreement used to be generally that you would spend more time writing than I would have to spend reading and when the agreement changes the quality changes as well.

That's not it at all.

The problem with AI slop is that in general it's not worth reading, because to start off even the author deemed the topic wasn't even worth writing to begin with.

Worst, most AI slop blog spam was obviously not even reviewed by the bloggers who prompted them into existence. This means is also content that even the bloggers themselves felt wasn't worth reading at all.

But somehow they expect us to invest our time reading AI slop?

[bombcar]

Think how much LinkedIn slop wasn’t worth reading BEFORE AI; but at least some effort had to be put in. Now none is needed.

[pydry]

Or rather because they are always just of poor quality

[ungovernableCat]

If you had a time machine and you could republish a claude generated article about some interesting tech topic to 2010 I'm sure it would get ok engagement.

To me the issue is that everything becomes written in the same style pattern. I don't know why but if I spot it I wince internally and immediately skim the post or just outright stop reading. A large part of it is overexposure. For me it's made browsing the internet quite uncomfortable since it's unavoidable.

I mainly spend my online time in group chats these days.

[27183]

> ok engagement

That is such a miserably low bar

[pmarreck]

Define “quality”

(I’m pretty sure both you and I would “know it when we see it”, but…)

[lproven]

There was an entire book about that 50 years ago, and last month someone brought that thinking and analysis to the LLM-bot issue:

EDIT: this is the one I was thinking of...

https://sinclairtarget.com/blog/2026/06/01/quality-in-the-ag...

But this is relevant too...

https://bersas.substack.com/p/zen-and-the-dance-of-generativ...

[mbesto]

You can't unilaterally say all AI blogs are low quality. That's bullshit.

> would spend more time writing than I would have to spend reading

To your point however, the reason people don't like AI generated blogs is because there is a explicit recognition that the author of the blog lacked effort. There is a visceral response for the reader about the social contract "if you didn't spend as much time as I did why should I care about what's written here", it's however NOT that the quality is inherently poor (but perhaps one my insinuate that notion).

[deaton]

> You can't unilaterally say all AI blogs are low quality. That's bullshit.

I can and I will say that, if not only for the fact that, in the eyes of many people, AI and the botched launch of the past few years has incurred a great amount of distrust. But, going further, the vast majority of AI-generated blog posts are lower quality, because they lack human thought and effort behind them. I'm not saying AI is completely incapable of writing what a human can, but we can't relate to how it "thinks", if you can call it that, and if someone is going to put enough effort into curating an AI's output and coaching it to make it output something that is really of a high quality... they just put in all of the human effort it would have taken for them to write it themselves.

I also don't think this should be construed to mean that human-written posts are universally good. AI slop is just the latest and most farmable iteration of a long history of badly written and poorly thought out content on the internet. In some ways the average AI blog is probably more coherent than any flat earth blog.

I also think that that visceral disgust at consuming AI generated work points to something else. We are all still trying to grapple with the ethical boundaries of what is okay and what isn't okay to do with AI, but I think most people feel deceived when they find out they're watching an AI video, or reading an AI blog post, and rather than assuming people are wrong to feel that way, we should consider that their feelings on the matter do matter. Nobody wants to be fed algorithmically optimized fake slop by YouTube, and that's okay.

[andsoitis]

> You can't unilaterally say all AI blogs are low quality. That's bullshit.

Can you point us to a single blog (or even blog post) that is generated by AI that is excellent quality? Or even average quality?

[asdf88990]

> You can't unilaterally say all AI blogs are low quality. That's bullshit.

Quality is a bit nuanced but that quality of AI generated content is subpar to expert output is fairly established.

The implications of this ever becoming false amounts to GAI.

[asdf88990]

AGI*

[lproven]

> You can't unilaterally say all AI blogs are low quality. That's bullshit.

Watch me.

I absolutely can and do. And pictures, and videos, and picture posts, and comics, and all image-format content.

If it wasn't worth creating with someone's brain, it is never ever worth my time to consume the prompt output.

Send me the prompt. I might look, if I'm bored.