Have you ever tried to write something with an AI tool like ChatGPT? Plenty of people rely on AI tools both in their personal lives and at work. Still, even with solid prompting, AI-generated content can sound a little … off.
Sure, factual errors and clumsy phrasing are pretty easy to spot, but sometimes smaller choices, like punctuation, can contribute to readers’ feelings of unease.
In fact, the humble em dash (—) is at the center of the AI content debate. It’s the darling of some writers who swear it can make prose feel more alive. But it also draws the ire of editors (and now, AI critics) who say it’s overused and the digital equivalent of a nervous tic.
To complicate matters, AI is known for its rampant use of the em dash. The problem isn’t whether you or ChatGPT uses an em dash. What really matters is that the writer understands their audience.
So, why is AI writing such a problem for brands? And why is the em dash such a big deal, anyway? Learn why this change is happening, plus actionable ways you can make copy more reader-aligned, regardless of who wrote it or how many em dashes it uses.
AI Writing: Friend or Foe?
There’s no sugarcoating it: AI content is everywhere, whether brands want to admit it or not. A study by Ahrefs found that 74% of new webpages include at least some AI content. Not to alarm you, but that means most of what you read online today has at least been touched, if not outright built by, AI writing tools.
For AI’s defenders, that’s a win. This camp believes AI blog writing is a valid tool that helps human writers create better content, faster. It eliminates spelling errors, speeds up brainstorming, and gives writers a base to work from. In the right hands, it’s a powerful tool that can free up time for strategy and creativity.
But AI critics see something else. They believe AI writing is formulaic and hollow. To be fair, this isn’t untrue. A lot of AI writing tools generate text, but they somehow manage to say nothing of substance in the process.
You might think readers can’t tell what’s AI, but studies indicate that your audience is pretty savvy. In a survey of 2,000 people, 55% of U.S. participants successfully identified AI-generated content. To make matters worse, a separate study found that 55% of people are uncomfortable reading AI-generated content, and say it makes them feel distrustful of a brand.
Oddly enough, much of the rage against AI has zeroed in on a bizarre scapegoat: the em dash.
The Curious Case of the Em Dash
AI critics say that the overuse of an em dash is a classic sign of AI writing. Sometimes that’s true, but it’s definitely not the full story.
The em dash has been around far longer than ChatGPT. It comes from the age of woodblock printing, where printers would create a dash the same length as the “m” woodblock. The use of the em dash was popular in the days of woodblock printing, but it nearly disappeared when typewriters entered the scene. Their simplified keyboards didn’t include an em dash, so it just wasn’t practical for writers to fashion their own.
Of course, em dashes are back today, with a vengeance. Even before AI, editors rolled their eyes at the em dash. It’s always been a bit of a chaos agent in grammar. People use it when they’re not sure how semicolons work, when they’re patching up comma splices, or when they just can’t figure out where to end a sentence.
And yes, the em dash has a dedicated cult following. Em dash enthusiasts use the punctuation as a stylistic choice to draw the reader’s attention with an intentional pause. Used sparingly, it’s a tool to add flourish. Overdo it, though, and you drift into Emily Dickinson territory (the poet was basically on a first-name basis with the em dash).
Em dashes were already heavily debated, but now that ChatGPT drops them into nearly every paragraph, it seems like they’re everywhere—and that’s because they are. A study found that one-third of all AI-generated sentences have em dashes; in human writing, they happen once a paragraph or less. A separate study found that the use of em dashes doubled from 2021 to 2025. No other punctuation saw that kind of spike.
Why on earth does AI love em dashes so much?
The logic is pretty simple. Algorithms rely on em dashes because they appear frequently in digitized books, so the model learns to replicate them. Plus, it’s computationally cheaper: why wrestle with commas, conjunctions, or sentence rewrites when an em dash stitches things together fast?
To put it bluntly, AI uses em dashes not to make better prose, but because it’s lazy.
When Style Meets Context
Em dashes and ChatGPT have their lovers and haters, but they both have their proper time and place.
Em dashes, for one, are popular in classic literature and poetry. Nineteenth-century novels were often dense blocks of text with few visual breaks. A sudden dash could jolt a reader awake and offer some breathing room. In that setting, the flourish worked. But in a 1,000-word blog post? Not so much.
What works for a novel doesn’t translate into the conversational writing style used in business blogs and marketing content. Overusing em dashes in blogs, LinkedIn copy, or your website makes the whole thing feel off. Real people don’t talk in em dashes. We pause with commas, ramble with ellipses, or break things up with short sentences. When AI leans too hard on the em dash, it breaks the illusion that you’re speaking with a human voice.
In other words, write in the vernacular. Remember when (I don’t and you probably don’t, but history remembers) books were only printed in latin? There weren’t many readers. Then someone published a book in the vernacular and it was an instant classic.
Context is everything. Writing style and tone evolve with the times. The way readers absorbed text in the 1800s isn’t the way we skim content on mobile today. Audiences expect direct, approachable, reader-friendly language.
AI’s Blind Spots
So, why does AI keep tripping over the same punctuation habits? Simple: it defaults to its training data. Large language models don’t “learn” the way humans do. They run statistical matches across mountains of text to predict which word or punctuation mark should come next. That’s why so much AI writing sounds like something you could just as easily find with a Google search.
AI is blind to context. Unless you prompt it with surgical precision, AI tends to insert stylistic quirks, like the overuse of em dashes, where they don’t belong. That’s what gives content that eerie, off-kilter “AI voice.”
But let’s be crystal clear: AI isn’t the villain.
It’s a tool, and a useful one at that. Like a computer, it does what you tell it to do. The problem comes when people accept whatever it spits out without checking for writing style and tone that fits their audience.
Bringing Back the Human Voice
Pandora’s box is open now, and we can’t go back to a pre-AI era. What we can do, though, is redefine our relationship with this technology.
As the human editor, you’re responsible for sharing content that makes a difference for your audience. And yes, that means you can use AI if you want, but you need to use it responsibly. Follow these best practices to write helpful, human-focused copy that actually helps your readers.
Fact-Check Everything
Eighty-two percent of AI users say they’re skeptical of its outputs, but just eight percent actually check the AI’s sources. If you’re publishing brand content, you can’t be in that 8%. Be the adult in the room: validate numbers, names, and dates before posting AI content.
Edit Content Ruthlessly
Never copy-paste something directly from ChatGPT onto your website or LinkedIn profile. AI can help you generate a rough first draft, but it’s on you to edit it into something worth sharing.
For starters, go through the draft and remove any AI quirks, like em dashes. You don’t need to remove them completely, but you should ration them. As a rule of thumb, aim for ~2 uses per 1,000 words in business copy.
If a dash creates meaning or emotion you can’t get with a comma or period, keep it. If it’s just papering over a messy sentence, rewrite it. Tools like Grammarly and Hemingway will help you make your pieces grammatically stronger and refine your writing skills.
Read Your Content Aloud
You should do this whether you wrote the content yourself or with ChatGPT. Reading content aloud before posting it will pinpoint any bits that sound awkward or robotic.
Read the draft yourself or use a read-aloud tool. In Google Docs, go to Tools → Audio → Listen to tab or insert an audio button via Insert → Audio buttons → Listen to tab to hear how it sounds. Since conversational writing is so important in business copy, this is a must for writing in a proper cadence.
Give AI Tools More Context
Grammar matters, but you need to make valuable content aligned with your target audience’s pain points. AI can write something that sounds good, but without additional information, the content just won’t get views, clicks, or conversions. And isn’t that the whole point of writing in the first place?
Before telling the AI to “Write a piece about X,” give it context first. It needs information on:
- The target audience
- Their pain points
- Your brand’s tone of voice
- Target reading grade level
- Length
- Calls to action
ChatGPT will do a much better job of writing content with this information. As always, double-check everything it generates. If anything feels off, tell the chatbot to rework the copy. It will save this information for future prompts, so the content should get better over time.
Blame Your Copy, Not the Em Dash
If you happen to love the em dash, don’t feel like you need to cut it out of your toolbox just to “not seem AI.” A sentence full of em dashes can still be engaging if it actually conveys something meaningful.
Em dashes aren’t the villain here. The real issue is context. When you use em dashes with intent, they add personality and rhythm to your copy. But if you (or ChatGPT) use it too much, it’s a grammatical crutch. The same goes for AI writing. Dismissing it outright or blindly worshiping it misses the point.
Context is content. Tools only work when the writer uses them with purpose. AI blog writing can spark ideas, speed up production, and make the job easier. But it can’t replace judgment. That’s on you.
It’s time to stop publishing filler content with AI tools. Map a content journey that wins attention and drives revenue: Work with Odyyc to start doing what works.
Leave a Reply