Are You Offended by the Em Dash (—)? AI Writing and Authenticity

•

How a tiny punctuation mark became a symbol of AI-generated writing, and what it reveals about trust, authenticity and human responsibility.

The Character That Makes You Suspicious

Have you noticed it? You’re reading an email from a colleague, a document, or perhaps a personal message. Everything looks beautifully written. The sentences flow naturally, the grammar is perfect, and the tone is spot on. Then you spot it: —.

The em dash.

Suddenly, you find yourself wondering: Did this person actually write this? Or did they simply copy and paste something from Copilot? An even more uncomfortable question might follow: Did they bother reading it before sending it to me?

It’s fascinating how a punctuation mark has become an unofficial fingerprint of AI-generated content. The em dash has existed for centuries, but its frequent appearance in AI-assisted writing has given it an entirely new reputation.

The character itself isn’t the problem. What bothers us is what we suspect it represents: someone outsourcing not just their writing, but potentially their thinking.

How Did the Em Dash Become an AI Fingerprint?

Unlike the ordinary hyphen (-) or the en dash (–), the em dash (—) doesn’t have a dedicated key on most standard keyboards. You can produce it using shortcuts or automatic formatting, but it’s not something many people usually type in everyday emails or messages.

Generative AI, however, seems perfectly comfortable using it. Large language models frequently produce polished, structured prose, and the em dash fits naturally into that writing style. As AI-generated content becomes increasingly common, readers have started associating the character with machine-produced text.

But there’s a catch. An em dash is not proof of AI authorship. Professional writers used it long before ChatGPT existed. Many people also use AI to improve grammar or clarity without surrendering control of their ideas. Treating punctuation as an AI detector risks unfairly judging perfectly authentic human communication.

Still, perception matters. Once readers associate a particular writing style with AI, they may begin questioning the authenticity of messages containing it, even when those messages are entirely genuine.

Why Does AI-Generated Writing Sometimes Feel Less Genuine?

Consider two messages from a colleague thanking you for helping with an urgent issue.

Message A:

“Thanks for helping yesterday. Really appreciated you staying late to get everything sorted.”

Message B:

“I wanted to express my sincere appreciation for your exceptional support yesterday—your dedication, collaboration, and willingness to go above and beyond truly exemplify the values that make our team successful.”

Both communicate gratitude. The second is arguably more polished, yet many people would find the first more genuine. Why?

Communication is more than delivering information. It also conveys intention, attention and emotional investment. When someone personally writes a message, we often interpret that effort as a sign of sincerity. When the same message appears to have been generated instantly by AI, we may question how much thought went into it.

Research into AI-mediated communication has explored this phenomenon, sometimes described as an AI penalty, where perceived AI involvement can reduce judgments of authenticity or trustworthiness. The underlying issue is our expectation of human connection. We might appreciate AI helping someone prepare a technical report, but feel differently when it appears to have written a heartfelt thank-you, a personal apology or a message of sympathy.

The sender’s feelings may be entirely genuine. Unfortunately, the recipient can’t see those feelings, only the words on the screen.

The Bigger Question: Did the Author Actually Read It?

This is where my concern really lies.

Imagine receiving an important email requesting a decision. You paste it into ChatGPT and ask for a professional response. Within seconds, AI produces something impressive. You copy it, paste it into Outlook and click Send.

Efficient? Absolutely. But did you fully understand the original request? Did you verify the facts, review the commitments and consider whether the response actually represented your position?

There’s a fundamental difference between using AI to express your thinking and using AI as a substitute for thinking.

In the first case, AI is a writing assistant. In the second, the human becomes little more than a delivery mechanism for machine-generated content. Yet their name still appears at the bottom of the email, implying understanding, intention and responsibility.

There’s also a hidden productivity problem. AI makes producing lengthy, polished messages effortless, but that doesn’t necessarily make communication more efficient. A manager asks, “Can we deliver this by Friday?” AI generates four paragraphs about collaboration, priorities and commitment to excellence, without answering the question.

The sender saves time, while the recipient spends more time interpreting the response and asking for clarification. We’ve improved the efficiency of generating words without necessarily improving the efficiency of communicating meaning.

Sometimes the best answer is simply: “Yes, provided we finish testing by Wednesday.” No elaborate introduction. No unnecessary corporate language. Just a clear, considered response.

When Everyone Can Write Perfectly, What Makes Communication Valuable?

Generative AI is fundamentally changing the economics of written communication. Producing a professional-looking article, business proposal or email once required considerable effort. Today, almost anyone can generate polished content in seconds.

That’s an extraordinary achievement, particularly for people who struggle with writing or communicate in a second language. AI can help us express ideas more clearly and overcome barriers that previously limited participation.

But it also introduces a challenge: when polished writing becomes effortless, polish itself becomes a weaker signal of expertise, sincerity or understanding.

A beautifully written strategy paper doesn’t guarantee sound strategic thinking. An impressive job application doesn’t necessarily demonstrate genuine capability. An inspiring leadership message doesn’t prove personal conviction.

Increasingly, we need to judge communication by something deeper than presentation: the quality of its reasoning, the accuracy of its claims, the relevance of its message and the accountability of its author.

This is why I believe we need a new etiquette for AI-assisted communication. Not a set of rules banning certain words or punctuation marks, but a few simple expectations:

  • Own what you send. If your name appears on a message, you should understand and stand behind its contents.
  • Review meaning, not just grammar. Verify facts, commitments and context before sending.
  • Preserve your voice. AI should help you communicate more clearly, not make every message sound like the same corporate template.
  • Respect the recipient’s time. A shorter, direct response is often more valuable than several paragraphs of polished but unnecessary text.

Most importantly, we shouldn’t automatically accuse someone of using AI based on their writing style. Suspicion is not evidence, and a “—” alone tells us very little about the author’s actual involvement.

So, Am I Offended When I See “—”?

Not exactly. Yes, exactly!

I don’t object to people using AI to write emails, documents or even personal messages. I use AI myself, and I recognise its tremendous value as a thinking and writing partner. What concerns me is when AI becomes an excuse to stop thinking, reviewing or taking responsibility for what we communicate.

The em dash is innocent. It has simply become a convenient symbol of a much bigger cultural shift, one in which we can generate more content than ever before, but may invest less personal attention in what we send.

I’m not offended that someone used AI to write to me. I’m concerned when someone uses AI to avoid thinking about what they’re saying to me. Perhaps, instead of asking “Was this written by AI?”, we should ask a more meaningful question:

“Does the person who sent this actually own it?”

Because ultimately, authenticity isn’t determined by punctuation. It’s demonstrated through understanding, intention and accountability. And AI, regardless of how beautifully it writes, shouldn’t take that accountability away from us.