The Em-Dash Debacle

If you’ve been anywhere on the internet in the last year or so, you’re probably familiar with what I am calling the “Em-Dash Debacle”. In case you missed it, the core issue is this: LLMs tend to generate a lot of em-dashes in their text output. Most people do not write with proper em-dashes these days. People (at least english speaking people) write with a lot of dashes, but they tend to actually use the hyphen character when doing so.

I am a millennial and I grew up with computers. I was taught that it was grammatically correct to use dashes, not hyphens. Microsoft Word (and later, Google Docs) made this trivially easy because you could type two hyphens (–) and it would replace it with an em-dash character. You never had to memorize a keyboard shortcut for inserting the em-dash character. I rarely write in Word or Google Docs these days, but when I do, I still do that double-hyphen shortcut. It’s a habit that I apparently have not forgotten.

So why is em-dash usage suddenly a problem?

I think the main reason people are noticing the em-dash usage now is because most writing has moved away from legacy tools like Word. “Modern” writing tools like Notion, Obsidian, etc does not not do that charaacter substitution. Nor do websites like Twitter, where it has become quite obvious when text is being pasted from another place…for example, AI generated content. If you are writing directly in those apps or websites, the majority of folks are just using a hyphen where they would historically use an em-dash. These tools are not replacing the hyphens with em-dash characters. And when people do come across em-dashes in Twitter or whatever, the automatic assumption is that the text is not being written by a real person.

If you do properly use em-dashes in your writing, people assume you are using AI-generated content. The piling-on in places like Twitter/X has become so bad, that many people are not intentionally avoiding them in order to avoid being called out for sharing “AI slop.”