I've seen, via social media, a kind of backlash about the use of emdash in writings. Apparently, this is considered an automatic signal that AI was used to generate the writing/report/story, etc.
So, I went to an AI and asked about it. Here is the response:
The em dash (—) has become a psychological trigger for AI suspicion because Large Language Models (LLMs) use it with a statistical frequency and technical perfection that human writers rarely match. [1] While humans have used the em dash for centuries to add dramatic pauses or break up thoughts, AI uses it as a mechanical "structural crutch". This overuse has trained readers to subconsciously associate the punctuation mark with robotic writing. [1, 2, 3] 🤖 Why AI Models Love the Em Dash
- Mathematical Safety: AI models operate by predicting the next most likely word or character. An em dash is highly versatile; it allows the AI to pivot, inject context, or extend a sentence without committing to rigid grammar rules like those required by semicolons or colons. [1, 2, 3, 4, 5]
- Training Data Bias: LLMs are trained on vast amounts of high-quality journalism, academic papers, and literature—genres where professional editors frequently use em dashes. The AI mirrors these highly-rated sources. [, 2, 3]
- Hyper-Correct Formatting: Humans typing quickly in a business report usually take shortcuts, typing a simple hyphen (-) or a double hyphen (--). AI outputs the exact, elongated Unicode character (—) flawlessly every single time, making it stand out as unnervingly "perfect". [1, 2, 3]
- Compounding Clues: The em dash rarely acts alone. When a reader sees multiple em dashes paired with typical AI "buzzwords" (like delve, testament, tapestry, or pivotal), the brain instantly recognizes the pattern of machine-generated text
BTW, when using Claude, it's possible to set up a "rule" within the Settings > Instructions for Claude to not use emdash. Just thought you'd like to know.