It’s become a mark of LLM-generated text because of its common use in more professional typesetting situations - such as the large amounts of texts used to train LLMs - while being uncommon in everyday use due to less accessible on keyboards.
People (on Linux & MacOS) are fucking lazy. I’ve been using topographically correct quotes, apostrophes and dashes since forever. It takes a minute to learn which key + Opt/AltGr does em/en dashes.
Of course you’re right about Windows, which for some inexplicable reason ships with braindead useless keyboard layouts. I don’t blame its users for not seeking out better keyboard layouts, that’s much more involved than finding the right key combination.
I agree with you, but also, in casual text (at least in my experience), hyphens are used much more commonly in its place. I don’t even know how to type an em dash on a physical keyboard, and I’ve gotten too used to it to even bother on a touch-keyboard. While it’s not absolute proof of generated text, it is a red flag, imo.
I’ve been a fan of em-dashes for a long time, mainly to try and make sentences more coherent without overusing commas, semicolons, and whatever the “…” is called. It’s just another piece of the grammatical jigsaw that allows written language to kinda have a voice. But I also don’t use an em-dash like 5 times in a paragraph
I just gotta long-press the dash symbol on my phone to use it — like so. Now on a physical keyboard, I gotta really want to use it cause I can never remember the key combination. But I refuse to let AI ruin a perfectly good piece of punctuation for me
You do you. I do think it’s relatively rare (or at least not very widely noticed) to most people, so it’s going to be something of a struggle for the foreseeable future. Best of luck.
Dang, I was thinking “ellipsis” but then was unsure of myself and too lazy to search, so thank you! Honestly, idk anyone IRL who uses em-dahes casually, I just like using 'em (eyyy, a little em joke)
The fact you haven’t seen it and that it’s so common in LLM output just means there’s a huge amount of the internet you don’t look at. That could be a good or bad thing—depending on your perspective.
And in fairness, common things can still be a tell of some kind. The first time I saw a normal webpage rendered in Computer Modern was friggin’ surreal.
It looks like there’s been 1 emdash out of the ~306,100 characters you’ve typed on lemmy.
(I’m having some trouble with the API (I am spamming the hell out of it to get these numbers so I should probably stop…), I may be missing some of your comments.)
I don’t use them on my phone (android), but I use them whenever I type on a word processor. Word, LibreOffice, or any every other office suite most academics and scientists use (Google Docs being the exception, though idk anyone who uses Google Docs after undergrad) automatically converts punctuation with two dashes sans spaces–like this–to an em dash. Google Docs converts to an en dash. Not saying he’s using a word processor, just saying why they show up so much in longer forms of writing.
More relevant to this post: My wife uses an iPhone, and her phone automatically converts two hyphens sans spaces to an em dash. It’s completely possible he’s using an iPhone, which makes em dashes trivially easy to use.
It’s a good grammatical tool. Were my phone able to do the automatic conversion, I’d use it in basically every Lemmy post I write. Please don’t contribute to the perception that proper use of good punctuation means AI.
Don’t even have to long press hyphen — doubling a hyphen will get autocompleted to an em dash. I don’t even know how to type two hyphens in a row without iOS converting it to an em dash.
I never used it in windows (what kind of idiocy is having to use alt codes anyway?) But I use proper characters in Linux all the time as they’re only a compose sequence away.
em dash
Booo —
Y’all need to relax about a punctuation mark that Markdown does when you hit dash twice.
LLMs didn’t invent the em dash. It appears in the chatbot because it appeared in normal text.
It’s become a mark of LLM-generated text because of its common use in more professional typesetting situations - such as the large amounts of texts used to train LLMs - while being uncommon in everyday use due to less accessible on keyboards.
People (on Linux & MacOS) are fucking lazy. I’ve been using topographically correct quotes, apostrophes and dashes since forever. It takes a minute to learn which key + Opt/AltGr does em/en dashes.
Of course you’re right about Windows, which for some inexplicable reason ships with braindead useless keyboard layouts. I don’t blame its users for not seeking out better keyboard layouts, that’s much more involved than finding the right key combination.
I agree with you, but also, in casual text (at least in my experience), hyphens are used much more commonly in its place. I don’t even know how to type an em dash on a physical keyboard, and I’ve gotten too used to it to even bother on a touch-keyboard. While it’s not absolute proof of generated text, it is a red flag, imo.
I’ve been a fan of em-dashes for a long time, mainly to try and make sentences more coherent without overusing commas, semicolons, and whatever the “…” is called. It’s just another piece of the grammatical jigsaw that allows written language to kinda have a voice. But I also don’t use an em-dash like 5 times in a paragraph
I just gotta long-press the dash symbol on my phone to use it — like so. Now on a physical keyboard, I gotta really want to use it cause I can never remember the key combination. But I refuse to let AI ruin a perfectly good piece of punctuation for me
It’s obviously option-dash for en dashes and option-shift-dash for em dashes on MacOS keyboards.
You don’t even need to learn that, you just think what the most logical way to type them would be, try it once, are correct, and know how to do it.
I don’t have MacOS, but it’s nice they make that intuitive
Overuse commas ? You think they are going to run out. If you are overusing commas try some new sentences.
I’m traumatized by elementary school teachers telling me my sentences were too “choppy”. But verbose quickly becomes run-on… I still have to heal
You do you. I do think it’s relatively rare (or at least not very widely noticed) to most people, so it’s going to be something of a struggle for the foreseeable future. Best of luck.
Also, “…” is an ellipsis, fyi.
Dang, I was thinking “ellipsis” but then was unsure of myself and too lazy to search, so thank you! Honestly, idk anyone IRL who uses em-dahes casually, I just like using 'em (eyyy, a little em joke)
I had never seen it before the LLM surge. Although that might be a case of Baader Meinhof
I’ve been using en (not em) dashes for like 20 years.
They’re used more often in German typography, and I’ve only seen em dashes without spaces around them which is ugly, so I stuck with them.
Again – two dashes will do it. Click the document icon below this comment.
Fwiw, I believe that’s an en dash. Slightly smaller than an em dash (—), but bigger than a hyphen (-).
Again, I had never seen it before, regardless of the fact that it’s easy to type
The fact you haven’t seen it and that it’s so common in LLM output just means there’s a huge amount of the internet you don’t look at. That could be a good or bad thing—depending on your perspective.
And in fairness, common things can still be a tell of some kind. The first time I saw a normal webpage rendered in Computer Modern was friggin’ surreal.
Seriously, I’ve been accused of being AI just because I use em-dashes even though that’s never happened before a year ago. It’s really annoying.
Can you explain the significance?
Normal people don’t use it while typing. They’d use a normal dash, tops. AI loves using the em dash even where it doesn’t fit.
alt+0151 on windows
long press on hyphen on phone
we do absolutely use them
Of the ~3.3 million characters you have typed on lemmy, 133 of them have been an emdash.
What are you using to see that information?
I scraped all their comments directly from their user page with a selenium script, dumped them into a text file and opened that in Libre Office.
… I am elegance personified. Someone hire me to work on your codebase.
Can you make a leaderboard? We can see which of us is closest to LLM—I’d place myself quite high up.
Only if I get to vibecode the whole thing.
Hah, I see. Thanks.
OK bot
/s
Do me!
It looks like there’s been 1 emdash out of the ~306,100 characters you’ve typed on lemmy.
(I’m having some trouble with the API (I am spamming the hell out of it to get these numbers so I should probably stop…), I may be missing some of your comments.)
I don’t use them on my phone (android), but I use them whenever I type on a word processor. Word, LibreOffice, or any every other office suite most academics and scientists use (Google Docs being the exception, though idk anyone who uses Google Docs after undergrad) automatically converts punctuation with two dashes sans spaces–like this–to an em dash. Google Docs converts to an en dash. Not saying he’s using a word processor, just saying why they show up so much in longer forms of writing.
More relevant to this post: My wife uses an iPhone, and her phone automatically converts two hyphens sans spaces to an em dash. It’s completely possible he’s using an iPhone, which makes em dashes trivially easy to use.
It’s a good grammatical tool. Were my phone able to do the automatic conversion, I’d use it in basically every Lemmy post I write. Please don’t contribute to the perception that proper use of good punctuation means AI.
Of the 55,814 characters you have typed on lemmy, 1 has been an emdash.
Your single use of an emdash was also in a comment that appears to have been written by an AI.
🥵
Don’t even have to long press hyphen — doubling a hyphen will get autocompleted to an em dash. I don’t even know how to type two hyphens in a row without iOS converting it to an em dash.
not on android–or, so it seems
The triple hyphen is the markdown encoding for an emdash, (
---) in case anyone on android wants to start using them.Meh, I prefer long press — quicker that way.
That comment wasn’t really directed at people for whom long-press is an option.
Maybe they’re trying to say that we’re not normal? Idk, but either way — rude lol
I’ve used it long before LLM’s were a thing.
Just because most people don’t use them doesn’t mean “people don’t use them” — or else the LLM wouldn’t have put them there in the first place
I went through the trouble of learning the alt+0151 on windows and will certainly keep using it
I used to use it all of the time when I still had Windows and used alt codes
Some of us read books.
I never used it in windows (what kind of idiocy is having to use alt codes anyway?) But I use proper characters in Linux all the time as they’re only a compose sequence away.