Here’s Why You Should Keep Using Em Dashes (and other AI “tells”)
A few days back, I saw a comment on Threads that said “I just wrote 30 emails with Claude. Not an em dash in sight.”
And I don’t bother arguing with strangers on the internet, but I did think to myself, oh, friend. That’s not the flex you think it is.
First, because for most small businesses, emails to your list should take thought and time. They’re personal. We want to know you. And if you’re using an AI tool to “write” 30 emails in an afternoon… that’s duplicitous.(But that’s not the point of this post.)
But also, because of that em dash comment. “Not an em dash in sight.”
Look, I get that large language models like Gemini, ChatGPT, and Claude overuse em dashes.
But that doesn’t mean that em dashes are bad (or even that they’re a sign something was written via one of those AI tools.)
The reason LLMs love em dashes is because writers (actual, human writers) love em dashes.
Similarly, antithesis (“It’s not this. It’s that.”), 3-part written lists, bulleted lists, and other stylistic conventions are overused by AI because those models were trained on good writing, which exhibits those conventions.
It’s just that unlike skilled writers, tools like ChatGPT don’t have the discernment (yet) to decide how much is too much of a good thing.
Ane don’t be fooled — used stylistically, em dashes ARE a good thing.
Em dashes gently guide your reader to follow your pacing and interpretation.
Shakespeare was very specific about using different punctuation to instruct his actors on how to read a line. A semi-colon is a different kind of “stop” than a period, for example. College theatre programs have entire courses on analyzing the text of Shakepeare’s plays, including learning his not-so-secret punctuation code.
As a result, we’re able to interpret his writing similarly to the company of actors he worked with nearly 500 years ago!
Your readers may not be trained actors, but they do understand (often subconsciously) how to dance through your words almost entirely based on how you write. And a large part of that interpretation comes from the punctuation you use.
And while we’re very clear on interpreting periods as the full end of a thought, followed by a pause…
An em dash leaves just enough space for interpretation by your reader.
This punctuation allows your reader to become a collaborative interpreter of your writing. And isn’t that a beautiful thing?
They can indicate an interruption, a parenthetical phrase, a tonal shift, repetition for emphasis, a thoughtful pause…
Artistically, they’re elegantly sophisticated. And very versatile.
Even so, I use fewer em dashes now than I did in our pre-ChatGPT era.
I’m conscious of replacing them with ellipses and periods most of the time. Sometimes colons or semi-colons. (But semi-colons feel so stodgy, y’know?)
I don’t replace them with parentheses just because I already use a lot of parentheticals. But parentheses might be a good fit for you if you’re looking to reduce (please, not eliminate!) your em dashes.
If we continue to change our personal styles based on trying to be different from the generative AI, our writing actually gets less effective. And more importantly, when we respond reactively to separate ourselves from these elements we now associate with AI, our writing ironically becomes less human.
Imagine we remove em dashes from our writing entirely.
In 2026, em dashes are gone. Poof.
Gen AI will continue to be trained on the writing we create — now sans em dash — and will find new patterns to copy.
Maybe colons will be the next big “AI tell.” And then everyone will rush to remove colons from their work. And so on.
Aren’t we over chasing the next trend? Don’t we have to do that enough on Reels? Isn’t that exhausting?
And fundamentally, isn’t language worth preserving?
Yes, I know that language is dynamic; it changes in reaction to social evolution moment by moment.
But there’s something different happening now, something different than has ever happened before.
Not only are we adapting human language to suit the “requirements” of using robots to create for us (while attempting to convince others that we created this ourselves) instead of adapting as a response to human social change…
But also, the SPEED at which we would need to continually evolve in order to chase the robots means that we are having to react to changes that are happening exponentially faster than pre-AI.
Ultimately, I don’t want my writing to be mistaken for AI any more than you do.
But if you find em dashes useful, don’t skip them entirely. Keep using your favorite emojis, sets of 3 examples, and antitheticals.
And if you simply don’t care for them and they’ve never been part of your style, then don’t use them. In other words, you do you. Just don’t bend and twist YOU to play the robot game.
Our greatest strength lies in our humanity. And by changing our writing to avoid AI paranoia, we participate in the degeneration of our own creativity and potential.
Stay human out there, friend.
Want to learn some AI “tells” that AREN’T good writing?
Keep an eye out for a new blog soon. But not right away. Because unlike Claude, my brain can’t write 30 blogs in an afternoon. And I’m okay with that.
In the meantime, check out this Reel about AI “detectors” and follow me there for more copywriting, public speaking, TEDx, and business insights.