The Em Dash Enigma: Why AI Models Use Them and How You Can Too
The frequent appearance of em dashes (–) in AI-generated text has led to questions about their origin. The primary explanation is that AI models learn this stylistic feature from their vast training datasets. This data often includes professionally written content where em dashes are used deliberately for emphasis or to connect clauses—a practice that has grown in modern writing as an alternative to semi-colons or parentheses.
Auto-Correction and Training Data: The Key Link
A significant factor is the common writing workflow where authors type a simple hyphen (-) or two hyphens (--), and word processing software (like Microsoft Word or Google Docs) or markdown renderers automatically convert these into a proper em dash (–). AI models are typically trained on this final, rendered text. Consequently, they learn to output the em dash character directly, rather than the simpler hyphen(s) a human might type. One participant with a background in Technical Writing confirmed they've used em dashes for decades, often by typing two hyphens.
How to Type an Em Dash Yourself
For those curious about how to manually type em dashes, the discussion provided several handy shortcuts:
- macOS: Press
Option + Shift + - - iOS/iPadOS: Long-press the hyphen key (
-) and select the em dash from the popup menu. - Windows: Hold
Altand type0151on the numeric keypad.
Additionally, simply typing two hyphens (--) in a row is a widely recognized convention that many text editors and online platforms will automatically convert to an em dash.
A Tangential Observation: The Rise of Emojis
Beyond em dashes, the conversation also touched upon the increasing use of emojis by AI models. Some users find this trend, also prevalent on platforms like LinkedIn, to be excessive or even annoying in AI-generated responses, such as those from ChatGPT. This preference for cleaner output leads some to favor tools like Perplexity, which are noted for a more straightforward presentation style.