For the last few months, work has been going deep into AI “tools” and my fellow writers have talked about how readers think something was generated by an LLM if it contains em dashes. So imagine my delight at this takedown of that idea, absolutely riddled with em dashes in a beautiful way: Stop AI-Shaming Our Precious, Kindly Em Dashes—Please
“The idea—the terrible, mistaken idea—is that the use of em dashes in a piece of writing is a sign that the text was generated by AI. Some people have been saying this on, guess where, the internet. The implication is that human writers should avoid em dashes for fear of being mistaken for chatbots. No. Wrong. I am here to raze this implication to the very ground and salt the earth where it stood.”
As the author notes,
“If generative AI does have a predilection for em dashes, though, the reason is simply that many human writers use em dashes. Your chatbot also uses commas, just as human writers do. A chatbot does not have a consciousness. It does not “know” how to write, in any meaningful sense. It doesn’t have a style, because style requires thought, preference, and taste. A gen-AI chatbot is trained by scanning gargantuan amounts of text. Based on the patterns it detects in that text, it then assesses the probability that certain words and syntactic constructions will occur in proximity to one another.
[… ] In other words, it’s not accurate to say that the use of em dashes in a text is a sign that the text is AI-generated. It’s more accurate to say that the prevalence of em dashes in AI-generated text is a sign of how reliant the AI companies are on the human writers they want to replace.”