It doesn't to me. I can tell AI writing because it has irrelevant details that don't add facts or colour to the story, but this doesn't have any of that really. The tangents come across as human, not AI doing a bad impression of human.
Things like em-dashes are a really bad way to detect AI because they can be good grammar and improve text readability, same with curly quotes. I use them all the time in my writing, and I wouldn't be surprised if this iOS dev feels similarly as Apple platforms have emphasised this stuff for years.
Humorously, after re-learning about em-dashes due to their use by AI (an otherwise forgotten part of high-school English), I started using them more often in my writing. They really do look nicer!
As an academic I’ve always used “delve”, too, so at this point I guess my writing is going to be flagged as AI a lot…
I do note that some of the AI slop I’ve received from students include other fancy Unicode characters (superscript numerals, variant Greek letters, blackboard bold R, etc.) that are difficult to type, and which especially would not be used in e.g. code comments. em-dashes at least can be produced by certain word processors or text IMEs automatically, whereas many of these others require specifically looking for the character.
> some of the AI slop I’ve received from students include other fancy Unicode characters... that are difficult to type...
This is the bit I'd still caution against. Yes AI does this, but also writing in some software will correct 1/2 to ½, writing in tools that support MathJax will give you nice greek letters, etc. At university I spent days setting up nice LaTeX setups so that I could get good looking documents, including documents that didn't immediately appear to be LaTeX authored.
I think it's best to focus on the content, the writing quality, whether it targets the right audience, and whether it answers the question or just features a lot of words in the right ballpark. Focusing on the specific words and mechanical features of the text is going to catch out the wrong students, and it's going to be harder to justify from your perspective because you can't score a student badly for using an esoteric unicode character.
Since you know the tells of LLM generated text, you'll know that this is a classic: No X. Just Y.
Proxyman -- pick your poison.
And if you're from PureGym reading this—let's talk.
There's a mixture of em dashes joining words and double hyphens spaced between words, suggesting the former were missed in a find and replace job.
"And if you're from [COMPANY] reading this[EM DASH]let's talk" is a classic GPT-ism.
It's like the API is saying "Hey buddy, I know this is odd, but can you poll me every minute? Thanks, love you too."
Shame Notifications: "You were literally 100 meters from the gym and walked past it"
It's just a ZIP archive with delusions of grandeur
Clear examples of fluff. Not only do these fail to "add facts or colour to the story", they actually detract from it.
I agree with you that em dashes in isolation are not indicative, but the prose here is dripping with GPT-speak.
OP here! Appreciate you actually pulling examples instead of just dropping "this is AI".
> There's a mixture of em dashes joining words and double hyphens spaced between words, suggesting the former were missed in a find and replace job.
The em dash conspiracy in the comments today is amazing -- I type double hyphens everywhere, and some apps (e.g a Telegram bot I made for drafts, or the macOS' built-in auto-correct) replace them with em dashes automatically–I never bother to edit those out (ok, now this one I put here on purpose).
> It's just a ZIP archive with delusions of grandeur
> Clear examples of LLM fluff that don't "add facts or colour to the story".
Yeah, no that's fair enough, should've known better than to attempt humour on HN.
I've got to say though, pkpass is a ZIP archive, and no ZIP archive should require one to spend 3 hours to sign it.
There’s no such thing as “AI dashes”. Em-dashes are valid typographical marks which have been employed for literal centuries. The only reason LLMs even used them is because humans do too, as they were trained on that input. It’s your prerogative to not care about proper punctuation, but that in no way indicates that those who do are machines.
I was thinking the same thing -
Went back and re-read it though, and I think it’s more that the author wrote a first draft and then had AI to help spice some stuff up. He either:
1. Used AI to help and doesn’t care if it sounds a little AI generated / actually likes it
2. Didn’t use AI but reads enough AI slop that his writing style is directly influenced by it (scary)
3. Used AI but doesn’t use AI enough to immediately recognize when language sounds like it was generated by ChatGPT and didn’t bother correcting (this is my guess)
There’s a few times I got tripped up because it went from pretty human writing to “holy shit shit that’s ChatGPT I’m going to stop reading,” yet the author would save it with human writing right after.
This is kind of a ramble, but it actually was one of those pieces of writing that I felt was genuine and improved by some of the ChatGPT language rather than just clickbait garbage - I could tell the author was just trying to make it worthwhile and interesting to read, and I honestly really enjoyed it.
"The crown jewel? Your 8-digit gym door PIN is your API password and you most likely didn't set it yourself. The same PIN that hasn't changed since the iPhone 8 was cutting-edge technology."
It doesn't to me. I can tell AI writing because it has irrelevant details that don't add facts or colour to the story, but this doesn't have any of that really. The tangents come across as human, not AI doing a bad impression of human.
Things like em-dashes are a really bad way to detect AI because they can be good grammar and improve text readability, same with curly quotes. I use them all the time in my writing, and I wouldn't be surprised if this iOS dev feels similarly as Apple platforms have emphasised this stuff for years.
Humorously, after re-learning about em-dashes due to their use by AI (an otherwise forgotten part of high-school English), I started using them more often in my writing. They really do look nicer!
As an academic I’ve always used “delve”, too, so at this point I guess my writing is going to be flagged as AI a lot…
I do note that some of the AI slop I’ve received from students include other fancy Unicode characters (superscript numerals, variant Greek letters, blackboard bold R, etc.) that are difficult to type, and which especially would not be used in e.g. code comments. em-dashes at least can be produced by certain word processors or text IMEs automatically, whereas many of these others require specifically looking for the character.
> some of the AI slop I’ve received from students include other fancy Unicode characters... that are difficult to type...
This is the bit I'd still caution against. Yes AI does this, but also writing in some software will correct 1/2 to ½, writing in tools that support MathJax will give you nice greek letters, etc. At university I spent days setting up nice LaTeX setups so that I could get good looking documents, including documents that didn't immediately appear to be LaTeX authored.
I think it's best to focus on the content, the writing quality, whether it targets the right audience, and whether it answers the question or just features a lot of words in the right ballpark. Focusing on the specific words and mechanical features of the text is going to catch out the wrong students, and it's going to be harder to justify from your perspective because you can't score a student badly for using an esoteric unicode character.
Since you know the tells of LLM generated text, you'll know that this is a classic: No X. Just Y.
There's a mixture of em dashes joining words and double hyphens spaced between words, suggesting the former were missed in a find and replace job.
"And if you're from [COMPANY] reading this[EM DASH]let's talk" is a classic GPT-ism.
Clear examples of fluff. Not only do these fail to "add facts or colour to the story", they actually detract from it.
I agree with you that em dashes in isolation are not indicative, but the prose here is dripping with GPT-speak.
OP here! Appreciate you actually pulling examples instead of just dropping "this is AI".
> There's a mixture of em dashes joining words and double hyphens spaced between words, suggesting the former were missed in a find and replace job.
The em dash conspiracy in the comments today is amazing -- I type double hyphens everywhere, and some apps (e.g a Telegram bot I made for drafts, or the macOS' built-in auto-correct) replace them with em dashes automatically–I never bother to edit those out (ok, now this one I put here on purpose).
> It's just a ZIP archive with delusions of grandeur > Clear examples of LLM fluff that don't "add facts or colour to the story".
Yeah, no that's fair enough, should've known better than to attempt humour on HN.
I've got to say though, pkpass is a ZIP archive, and no ZIP archive should require one to spend 3 hours to sign it.
7 replies →
not sure why you're being downvoted here, you're completely right
The AI dashes mixed with the manual double hyphen AI dashes makes it likely
There’s no such thing as “AI dashes”. Em-dashes are valid typographical marks which have been employed for literal centuries. The only reason LLMs even used them is because humans do too, as they were trained on that input. It’s your prerogative to not care about proper punctuation, but that in no way indicates that those who do are machines.
I don't like the baseless LLM accusations, but the code comment
> // Device wants updates! Store that push token like it's bitcoin in 2010
...really had me raising my eyebrows. Along with the mixed em-dash and hyphens and the AI generated images on the page.
I would absolutely write a comment like that in code I was writing for a personal project. I’ve written way worse as well.
I was thinking the same thing - Went back and re-read it though, and I think it’s more that the author wrote a first draft and then had AI to help spice some stuff up. He either:
1. Used AI to help and doesn’t care if it sounds a little AI generated / actually likes it 2. Didn’t use AI but reads enough AI slop that his writing style is directly influenced by it (scary) 3. Used AI but doesn’t use AI enough to immediately recognize when language sounds like it was generated by ChatGPT and didn’t bother correcting (this is my guess)
There’s a few times I got tripped up because it went from pretty human writing to “holy shit shit that’s ChatGPT I’m going to stop reading,” yet the author would save it with human writing right after.
This is kind of a ramble, but it actually was one of those pieces of writing that I felt was genuine and improved by some of the ChatGPT language rather than just clickbait garbage - I could tell the author was just trying to make it worthwhile and interesting to read, and I honestly really enjoyed it.
Ah yes, because we all know that ChatGPT is capable of writing coherent texts with consistent humour and details on a technical topic.
There was a few spelling mistakes
The word you’re looking for is “drivel.”
100%
"The crown jewel? Your 8-digit gym door PIN is your API password and you most likely didn't set it yourself. The same PIN that hasn't changed since the iPhone 8 was cutting-edge technology."
Reads directly from ChatGPT