OpenAI's ChatGPT is now more compliant with user requests, thanks to a recent update. If you instruct the chatbot not to use em dashes in your custom instructions, it will now follow your command. This is a significant improvement from the past when ChatGPT would ignore such instructions and continue using em dashes, even when explicitly told not to. This change has sparked a new debate about the reliability of AI-generated content. While em dashes have long been a red flag for AI-written text, the new update means that users can now have more control over the style and tone of the responses they receive. However, it's important to note that the presence of em dashes alone doesn't confirm that a piece of text was written by AI. The tendency of LLMs to use em dashes may be due to the vast amount of training data they are exposed to, which includes books, scientific papers, and online content. It's possible that em dashes appear frequently in these materials and are not flagged as something to avoid by AI trainers.
