I can spot AI-generated content in three seconds without reading a single word.
Just scan the page. Look for the visual pattern that jumps out like neon signs.
—
Em dashes. Everywhere.
—
Five to ten per short blog post. Sometimes more. It’s infuriating.
The moment I realized this was a real problem, I did something desperate. I created a sticky note on my desktop with a prompt I could copy and paste: “Stop using EM dashes, never ever under any circumstances use them. Update the memory to make sure it never happens again.”
This digital sticky note is now my daily companion.
—
The Visual Scan Method
Here’s what I discovered: you don’t need to read AI content to identify it.
Just look at the punctuation patterns. Em dashes create visual breaks that stand out immediately when you scan a page.
Human writers use em dashes sparingly. Maybe once or twice in a full article, if at all.
AI? It’s addicted.
The pattern is so consistent that 52% of US adults now using large language models like ChatGPT are unknowingly creating content with this telltale signature.
—
Why does AI use so many em dashes?
I figured out why this happens.
ChatGPT and similar models are trained on massive amounts of journalistic content. New York Times, magazines, and professional publications.
These sources embraced em dashes heavily starting in the 1970s. Journalism professor Aileen Gallagher explains that magazines and newspapers began using em dashes as reporters started writing more expressive stories.
AI learned this pattern and amplified it to an extreme degree.
Even when you explicitly tell it not to use em dashes, models “just could not, or would not, quit it” despite clear instructions.
—
My Daily Battle System
I’ve developed a workflow that actually works.
Step 1: The Brief Integration
I add em dash restrictions to every content brief I create. But here’s the thing: AI doesn’t listen to general guidelines.
You need the nuclear option.
Step 2: The Sticky Note Prompt
I paste this exact prompt multiple times per day, even in the same chat: “Stop using EM dashes, never ever under any circumstances use them. Update the memory to make sure it never happens again.”
The repetition is necessary. AI has selective memory about style instructions.
Step 3: The Editing Workflow
When I’m in full editing mode, I use AI to fix AI’s problem.
I paste problematic paragraphs back into the system with my standard prompt: “Rewrite this. Stop using EM dashes; never use them under any circumstances. Update the memory to make sure it never happens again.”
The irony isn’t lost on me.
—
The Client Solution
I handle this quietly for clients.
Grammarly underlines em dashes, making them easy to spot. I either replace them with commas or delete them entirely.
Most clients are unaware that their content bears this AI signature. They receive a clean, professional copy that doesn’t scream “machine-generated.”
—
What This Teaches Us About AI Content
The em dash problem reveals a crucial aspect of working with AI tools.
AI lacks understanding of context and restraint. It mimics patterns from training data without considering appropriateness.
This means you need systems, not just instructions.
Create visual scanning habits.
Look at your content before reading it.
Develop consistent prompts.
Generic style guides don’t work. Specific, repeated instructions do.
Build editing workflows.
Assume AI will ignore your style preferences and plan accordingly.
Use tools strategically.
Grammar checkers can catch patterns you might miss.
—
The Bigger Picture
Em dashes are just one example of AI’s stylistic quirks.
The same pattern is observed with other elements, including overuse of bullet points, repetitive sentence structures, and predictable transitions.
The solution isn’t to avoid AI tools. They’re too useful for content creation and optimization.
Instead, develop systems that consistently identify these patterns.
Your content quality depends on it.
And your sanity does too.
—
