You've probably noticed it without being able to name it. That particular rhythm in AI-generated content where thoughts get connected with a long dash. It creates a certain cadence. A pause that feels sophisticated at first, then starts to feel mechanical once you see it everywhere.
I'm talking about the em dash. And it's become the most reliable indicator that content was written by (or heavily assisted by) AI.
The Pattern Recognition Problem
Here's what happens when you use AI for content creation: the language models have learned that em dashes create a sense of thoughtful pause. They add what feels like intellectual weight. So the AI uses them. A lot.
The problem isn't that em dashes are bad punctuation. Professional writers use them effectively all the time. The problem is frequency and pattern.
When every other sentence contains an em dash creating the same rhythmic pause, readers start to notice. Even if they can't articulate what they're noticing, something feels off. The writing has a sameness to it. A predictable cadence that signals "this wasn't written by a human thinking through these ideas."
Why This Matters for Your Business
If you're using AI to help create content (and you probably should be), you need to know this tell exists. Because your readers, customers, and clients are developing the same pattern recognition.
They may not consciously think "this was written by AI." But they'll feel something. A lack of authenticity. A sense that no one really sat down and thought through these ideas for them specifically.
And in a world where AI-generated content is flooding every channel, authenticity becomes a competitive advantage.
The Fix: Voice Calibration
The solution isn't to avoid AI tools. It's to calibrate them properly.
In my work developing voice frameworks for AI collaboration, eliminating em dash overuse is literally the first thing I address. It's Phase 1 of voice calibration because it's the most obvious tell and the easiest to fix.
Practical steps:
- Audit your existing content. Search for em dashes in your recent AI-assisted work. Count them. If you're seeing more than one or two per 500 words, you have a pattern problem.
- Establish explicit guidelines. When working with AI, include specific instructions about punctuation variety. "Avoid em dashes. Use periods, commas, colons, and semicolons to create rhythm variety."
- Read aloud. The em dash pattern becomes obvious when you hear it. That pause, that pause, that pause. If you're hearing the same rhythm repeatedly, revise.
- Develop alternatives. Most em dash usage can be replaced with other punctuation or restructured sentences entirely. "The solution isn't complicated, it's just overlooked" works as well as "The solution isn't complicated (it's just overlooked)."
Beyond Em Dashes
This is just one example of AI writing tells. Others include:
- Overuse of "straightforward," "I'd be happy to," and similar phrases
- Generic transitions: "That said," "It's worth noting," "In essence"
- Lists that always have exactly 3 or 5 items
- Conclusions that start with "In conclusion" or "To summarize"
- Hedge words clustered together: "may," "might," "could potentially"
The em dash is just the most visible symptom of a larger issue: AI defaults to certain patterns, and those patterns become recognizable with exposure.
The Real Opportunity
Here's what most people miss: AI writing tells aren't just problems to fix. They're opportunities to differentiate.
When everyone's AI-assisted content sounds the same, genuinely calibrated content stands out. Not because it hides AI involvement, but because it demonstrates that someone cared enough to ensure the output reflects real thinking.
Your readers can tell the difference between "AI wrote this and no one reviewed it" and "AI helped with this, but a human shaped it with intention."
The em dash epidemic is your chance to be in the second category.
Quick diagnostic: Copy your last AI-assisted piece into a document. Use Find to search for "—" (the em dash character). If you find more than 2-3 in a typical blog post, you've got calibration work to do.
Moving Forward
The goal isn't AI avoidance. It's AI collaboration that produces genuinely useful, authentically voiced content.
Start with the em dash. It's the easiest tell to spot and fix. Then work outward to other patterns. Build voice calibration into your AI workflows from the beginning rather than trying to edit it in afterward.
Your readers will notice the difference. They may not know why your content feels more authentic. But they'll feel it. And that feeling translates into trust, engagement, and ultimately, business results.