More Info About Personalised Private Videos: https://www.personal.thebluecats.com.sg/ More info about em dashes: https://www.grammarly.com/blog/punctuation-capitalization/why-you-should-love-the-em-dash/ Many regular AI users have noticed a peculiar trend: AI-generated text, especially from ChatGPT, tends to overuse the em dash. This has become so distinct that some people can spot AI-generated writing just by seeing frequent em dashes. But what exactly are em dashes? In English, there are three kinds of dashes—hyphens, en dashes, and em dashes. Hyphens link words (like “part-time”), en dashes represent ranges (like “Monday–Friday”), and em dashes are the most versatile. Em dashes can replace commas, colons, semicolons, or even brackets, giving a stronger impact to a sentence or shifting its tone abruptly. Because em dashes are powerful and expressive, they’re popular in writing that wants to convey sudden shifts or emphasis. However, using them isn’t straightforward; there’s no dedicated em dash key on keyboards, which means most people don’t use them often. On Windows, you’d need to press ALT + 0151, and on Mac, Option + Shift + Hyphen. This makes their overuse even more suspicious, especially when human writers rarely bother with them. A recent podcast by two Gen Zers went viral after calling it the “ChatGPT hyphen,” further cementing the em dash as an unofficial AI fingerprint. So, why is ChatGPT in particular obsessed with it? According to user experiences and Reddit discussions dating back to late 2024, models from OpenAI, especially ChatGPT and Microsoft’s GPT-powered tools, began churning out texts littered with em dashes. In contrast, other AI models like Google’s Gemini or Meta’s LLaMA hardly used any. It’s likely that as ChatGPT continued training, it found em dashes very useful—unburdened by the physical effort humans face in typing them, the AI just used them whenever suitable. Compared to human writing, this habit became glaringly obvious. Ultimately, it’s not that em dashes are inherently “AI-sounding”; it’s that AI now uses them more than we do. As a result, they’ve become a red flag, even when a human writer might be using them naturally. Over time, other models may adopt this quirk too, depending on how they’re trained. But one thing AI still can’t replicate well? Singlish lah. So if you want to hide from AI detectors, maybe start writing like you’re talking to your kopi kaki at the hawker centre.