[eterm]

It isn't "clearly human-written" at all, the entire blog looks like LLM output, right from the very first post.

I'm not witch-hunting, there are just a lot of witches.

[efilife]

I just went through some of the posts and you are right. It's very suspicious, but I would say it's right at the edge of being plausibly written by a human. If it's LLM, then it's the first one I'm aware of that got me this good. I am usually the first one to point out that something reeks of LLM writing here (which I'm kinda ashamed of, considering how much I've been doing this).

Tbh the whole smolweb concept by this person seemed kinda weird right when I discovered it was a thing. It seems to not really be a thing but the person is really trying to convince you that it is

[anotherevan]

“Ever since I learned about confirmation bias I’ve been seeing it everywhere!”

The raging, “I don’t like this, so it must have been written by an LLM!” comments on HN have gotten so tiresome that I find when I see them I just down-vote them and move to the next thread. (Most of the time. You’re witches comment captured my attention and prompted a response. Well done — the comment must have been written by a human.)

— No tokens were harmed in the production of this comment. —

[eterm]

There's no rage, and often I like or agree with the article, but it's still written by LLM.

That doesn't fill me with rage, just a sadness that people aren't sharing their own work and aren't using their own voice.

[surgical_fire]

Honestly, despite the regular "not X, just Y" constructs, I actually think it was written by a human. Or at the very least mostly written by a human.

Something about how the argument and rationale builds up does not scream LLM to me.

[eterm]

You may be sleeping on just how good LLMs have got at writing blog-posts.

Go ahead and ask your favourite one this:

> Can you draft a blog post titled, "All my clients wanted a carousel, now it's an AI chatbot!"

> Don't search the web, just go with vibes.

I did, and this was the result: https://richardcocks.github.io/chum/blogexample.html

Okay, not quite there, very much more obviously LLM than the OP, but a bit of tweaking, some feedback to drop the headings and the table, and:

https://richardcocks.github.io/chum/blogexample2.html

And that's with zero blog-writing "skills", with no memories, a fresh incognito session and only the title to prompt.

Complete with call-out:

> The feature was never really about the users. It was about the client feeling like they were keeping up. The technology changes. The psychology doesn't.

Complete with the horse-shit, "Honest dispatches from a decade in the web trenches"

[surgical_fire]

You may have a point. The example you posted was a bit more obvious to be the work of a LLM, but not by far.

The interesting bit is that I don't really care about the subject matter. I was browsing the comments section and the discussion of whether the blog post was AI generated picked my interest, so I tried my hand at reading to see if I agreed or not.

I wonder what to make of this. Once the lines between LLM written and human written are blurred, what is the outcome?

In some scenarios I think it's alright; I honestly don't care if a tutorial on how to set up an application is AI generated, as long as it is correct. Hell, I routinely use LLM as a glorified web search for that exact thing.

Sometimes however it becomes pointless. An opinion piece being AI generated is little more than noise. What is even being attempted there? Raking in some adsense from page views? As long as people willingly engage with it, why stop?

The web has been for a long time a low-trust environment, and this exacerbates that. Why even bother to share an opinion.

[johannesberlin]

both examples reek of LLM writing though, none of these are good