OpenAI has introduced enhancements to ChatGPT, enabling users to reduce the overuse of em dashes in AI-generated text. While em dashes won’t be eliminated entirely, users can set preferences through custom instructions to limit their occurrence. OpenAI CEO Sam Altman confirmed this update, emphasizing that users can instruct ChatGPT not to use em dashes in their outputs. The em dash has recently become a clear indicator of AI-generated content, appearing frequently in various formats, prompting concerns about authenticity. Some attribute this tendency to the training data used for language models, which contained many 19th-century texts. Although OpenAI has improved ChatGPT’s handling of em dashes, they haven’t shared technical details on the adjustment. Users can customize their ChatGPT experience by navigating to the settings and inputting “Please do not use em dashes” in the custom instructions section. This personalization feature is available to all users, including those on the free tier.
Source link
