OpenAI has addressed the excessive use of em dashes in ChatGPT, a notable quirk that had become synonymous with AI-generated text. CEO Sam Altman announced that users can now set custom instructions to minimize or eliminate em dashes, marking a significant improvement with the release of the GPT-5.1 model. This change comes after years of user feedback and complaints about the chatbot’s inability to comply with specific formatting requests. The em dash, typically used for parenthetical information or sudden shifts in thought, became a telltale sign of AI writing, prompting concern among human writers. While theories exist about why AI models favor this punctuation—ranging from training data influences to reinforcement learning—the issue highlights deeper challenges in controlling AI behavior. Despite this update, reports suggest that some instances of em dash usage still occur, raising questions about the journey toward true artificial general intelligence and prompting discussions on AI’s limitations.
Source link
Share
Read more