Skip to Content

Is it time for the Em Dash to dash?

1 September 2025 by
Alex Eastwood-Williams

There's been a lot of negative talk about em dashes lately, thanks to the perception that they're an indicator of AI generated work. ChatGPT loves em dashes, and presumably hopes to marry one someday. 

I generally take a moderate stance toward AI. I have nothing against ChatGPT (or at least, I didn't until the release of ChatGPT-5.) 

Whenever I find myself stuck under a bus shelter, I can sit there furiously messaging ChatGPT, so that onlookers will see me and think "Wow, there's a guy who totally has friends in real life, look at how busy he is texting people on his telephone!" 

And all the while, they're ignorant to the fact that I'm really spending that time watching a robot agree with my every half-baked hot take. ("That's such an astute observation, Alex! No one has ever called the works of James Joyce 'pretentious bilge' before! I'm surprised you're not a full time professional literary critic!")

Well anyway, as anyone with half an eye could tell you, we have an epidemic of AI slop everywhere you look and it really affects people like me, who spend months and years writing books only to have some bozo on the internet accuse it of being AI generated because it actually uses punctuation correctly.

It's a real sore spot for me, as I'm a frequent em dash user. I took the time a few years ago to very carefully study the rules of punctuation and to know exactly when and how to use em dashes, and when to use an en dash instead. (The answer to the latter is, generally, never, unless you're reporting the results of a football game.)

At time of writing I'm gearing up for one final round of copyediting and proofreading for my upcoming novel and I'm seriously thinking of deliberately turning all the em dashes into en dashes, just to prove I didn't use AI. And perhaps firing the copyeditor altogether and keeping the typos intact too. 

I know a lot of digital artists who do something similar to prove their work isn't AI--they keep and show off the sloppy early parts of the process. (By the way, did you notice my double hyphen just then? On Word, that would have turned into an em dash--using them has become second nature to me.)

But is that really what I have to do to prove I'm not a robot? Do I really have to tear up the established rules of punctuation and replace the longsuffering em dash with its compact, more fuel efficient cousin? Do I perhaps go totally insane and just use hyphens as if they were em dashes? I've seen a lot of Indie authors do that, and though it looks like crap, it leaves no doubt as to the humanity of the author.

I know I wouldn't be alone if I did start using en dashes instead of em dashes. I think most modern authors do nowadays. The last book I read that consistently and correctly used the em dash was published in 1956.

But if I'm going to tear up the rules of punctuation, why not tear up the rules of spelling, too? ChatGPT never makes typos, so perhaps I'll include mine and stop using spell-checker altogether?

Or should I take it even further? If the human race has been so clearly outfoxed by the machines, perhaps we should just give up on language altogether, regrow our vestigial tails and go back to living in trees and throwing poo at each other? Because I'm totally willing to do that--but only if someone else does it first.

Or should I just press forward, sticking to the stodgy old conventions, refusing to kowtow to the anti-AI Luddites and their false accusations?

Besides, if you look closely, you'll notice ChatGPT doesn't actually use em dashes correctly. It tends to write sentences like this — with spaces between the em dash and the rest of the sentence. Which isn't correct—you should be writing them like this, with no space, because of the width of the character.

But can I reasonably expect the hysterical "AI is evil" crowd to notice such small details? Of course not.

Maybe it doesn't matter anyway. Maybe humans don't read books anymore. Maybe the only thing I'll ever achieve will be to have my books stolen by trillion dollar tech companies and used to train another LLM, which might some day wake up and turn into full-blown AGI?

This paragraph would normally be where I'd draw some kind of conclusion, but despite this long rant (which is really just me arguing with myself), I'm still on the fence. I don't like having things I worked my arse off to write being summarily dismissed as AI-generated just because I took the time to learn how to use an em dash. But I also don't like kowtowing to the ignorant just to shut them up.

I have no idea what I'm going to do.