alt Hacker NewsEvery time this comes up I have to ask, how do you know it is AI? Sure, sometimes there are strong tells, like if they forget to take off the "Do you want me to change anything" at the end.
But AI was trained on good writing. Good writing looks like AI. I've had to reduce my use of em dashes for example because they are an "AI tell". I used to like using the term "load bearing".
And the list gets longer every day. Eventually every mark of a good writer will be an AI tell.
I don't know how to resolve this. But I've been accused multiple times of copy/pasting AI when I did no such thing, and it's getting equally frustrating.
Replies
I know it's AI because the laziest coworker you know just posted 400 paragraphs of text on a minor topic.
It has that feel to it. Hard to describe. It has a very consistent way of writing that gives it away, even if it's rephrase from text that was previously written by a human.
Beyond that, most AI content is WAY WAY WAY WAY WAY longer than it needs to be to get its point across and is oversaturated with unneeded "mic drop" phrases
The problem with AI writing isn’t the tics and the tells. The problem with AI writing is that it’s bad.
The typical highly voted HN AI-written post has an annoying clickbait title, then jumps into a meandering chain of “so there I was” scene-setting, assertions of dubious reliability & relevance, and supposed supporting examples. All punctuated by short, unhelpfully dramatic sentences that would interrupt the high-level narrative, if only there were one.
I’m really curious what “good AI writing” you had in mind. My personal intuition is that for almost any topic, the HN LLM-generated post du jour has a far superior cousin sitting in the archive pre-2021.
[flagged]
Can you read something and tell when an author you like wrote it without seeing the attribution? I can. I think most of us could, because people use certain phrases or have a particular style. That doesn't make it bad.
AI writing isn't actually all that bad. But it has a very specific style, and there are MANY tells. I use Claude every single day, and it's immediately obvious to me when I'm reading something Claude wrote. "It's not X, it's Y" "That's the reframe" "The gap is" I could go on.
I too like using em dashes, and have been since I learned the keyboard shortcut over a decade ago. I've never been accused of posting ai-written text because of it.
AI is indeed trained on good writing, but it's also trained on bad writing. And it's post trained on specific things that are writing that aren't good or bad for reasons unrelated to style. What you're left with is a one-size-fits-all writing style. Doing math? Doing medicine? Doing anime waifu role play? It all starts from the same generic milquetoast style until you tell it not to. And to me (and apparently many other people) that's super easy to pick out.