What is it with cool projects using LLMs to write their blog posts? If i'm honest I'd rather see it the other way around, tasteful agent coding with a genuine write-up. Reading AI blog posts make me vomit. Why even start a blog post if you can't be fucked to write?
I think people write a rough draft and let AI clean it and fill in the gaps, which leads to all usual phrases. I suspect some people have become so reliant on this they're now desensitized to AI speak.
I've noticed it being heavily used in non-native and accesibility type situations.
To them, people who do not necessarily spend swathes of their day reading generated text, it does not sound any particular way other than "yes, this is what I would like to communicate to my perspective audience".
It's really becoming a sort of cultural thing in which one people (from your culture) are turned away and another people (from the author's culture) see it as a massive benefit.
To me, the bad thing about generated posts online aren't their existence. I've already spent a life clicking off of blogs and posts I don't like. Perhaps as much as 90% of the web is not made for me, and that has alway been the case. What I've never understood, which has also existed forever but less so here, is people who will spend their time adding nothing of value to a conversation adding only some complaint about something as trivial as the writing style. You know, the commenters who take their time to click on to a video and actually type out "this dumb you dumb" without offering any kind of constructive criticism or engaging in the conversation at all.
Like, you wouldn't do that in public, would you? Walk up to someone with a shirt you don't like, say that you don't like it, and walk away? "Why even go out in public if you can't be fucked to dress yourself?"
The use of AI in writing was deliberate - it is a bit of irony and live evidence. I'm a nerd, no writer in fact. Language and artistic wording is pretty hard for me. What I lack is skills. LLM's allowed me to skip the step and focus on taste. The argument is very well refined, please, dig into it. AI written but human designed. I wonder whether sharing the session with the stochastic parrot would be better, but no one wants to read a hundred messages reiterating the same paragraph because it is unsound or invalid.
Sharing the session would be better. What you got at the end is almost unreadable.
I had to paste it into Claude to try and dig the underlying argument out. The best version I got from that (and it's still not great) was:
Generation is free, so artifacts no longer prove
skill. Judgment can't be automated. A sub-1kB
page where every byte was judged proves a human
judged — a new status symbol.
I might have gone too far onto the fanciness side. Huge thanks for the feedback, I'm very deranged due to working entire days with AI generated text. I didn't notice it was incomprehensible to the normal eye. I will sanitize the conversation and add it to the post. It's quite a lot of sessions so it will take a bit.
Having AI write your blog post and then talking about taste is a fascinating combination
What is it with cool projects using LLMs to write their blog posts? If i'm honest I'd rather see it the other way around, tasteful agent coding with a genuine write-up. Reading AI blog posts make me vomit. Why even start a blog post if you can't be fucked to write?
I think people write a rough draft and let AI clean it and fill in the gaps, which leads to all usual phrases. I suspect some people have become so reliant on this they're now desensitized to AI speak.
Aversion to AI speak will turn into desensitization.
And desensitization will eventually turn into a preference for AI speak. (Perhaps not the HN crowd.. but definitely for the average Joe)
People prefer reading things in a tone/style that's comfortable and familiar to them. Writing in a way that mimics LLMs checks those boxes.
desensitised? Quite the opposite, I've become allergic.
I've noticed it being heavily used in non-native and accesibility type situations.
To them, people who do not necessarily spend swathes of their day reading generated text, it does not sound any particular way other than "yes, this is what I would like to communicate to my perspective audience".
It's really becoming a sort of cultural thing in which one people (from your culture) are turned away and another people (from the author's culture) see it as a massive benefit.
To me, the bad thing about generated posts online aren't their existence. I've already spent a life clicking off of blogs and posts I don't like. Perhaps as much as 90% of the web is not made for me, and that has alway been the case. What I've never understood, which has also existed forever but less so here, is people who will spend their time adding nothing of value to a conversation adding only some complaint about something as trivial as the writing style. You know, the commenters who take their time to click on to a video and actually type out "this dumb you dumb" without offering any kind of constructive criticism or engaging in the conversation at all.
Like, you wouldn't do that in public, would you? Walk up to someone with a shirt you don't like, say that you don't like it, and walk away? "Why even go out in public if you can't be fucked to dress yourself?"
posts like this only get on hn because nobody on hn actually reads them.
The use of AI in writing was deliberate - it is a bit of irony and live evidence. I'm a nerd, no writer in fact. Language and artistic wording is pretty hard for me. What I lack is skills. LLM's allowed me to skip the step and focus on taste. The argument is very well refined, please, dig into it. AI written but human designed. I wonder whether sharing the session with the stochastic parrot would be better, but no one wants to read a hundred messages reiterating the same paragraph because it is unsound or invalid.
Sharing the session would be better. What you got at the end is almost unreadable.
I had to paste it into Claude to try and dig the underlying argument out. The best version I got from that (and it's still not great) was:
I might have gone too far onto the fanciness side. Huge thanks for the feedback, I'm very deranged due to working entire days with AI generated text. I didn't notice it was incomprehensible to the normal eye. I will sanitize the conversation and add it to the post. It's quite a lot of sessions so it will take a bit.
> LLM's allowed me to skip the step and focus on taste
huh?
You make a load-bearing point.
Why even start a blog post if you can't be fucked to write?
If they're not going to write it, I'm not going to read it.
It sounds logical. The people who are attracted to one kind of low-effort output will be attracted to other kinds of low-effort output.
Yeah.
"The sequence is the whole story: overflow, then reduce, then squeeze, then stop when nothing can die."
"the decoupling is complete: skill is cheap, taste is scarce, and nobody trained to spot it."
"That is the handicap, and it is the whole game."
Given the nature of this project, why write it up with slop?
thanks for reading, the same reason I told in the post - lack of skills.
Sorry, am I going crazy or the text is LLM generated?
Is it intentional? Is it satire? Is it irony? Is it human generated but written as if it was written by Claude to be provocative?
There is something distinctly dystopian about someone using AI to write a blog post about how we sold out our taste to LLMs.