Researchers at OpenAI have developed a method for teaching AI models to express their uncertainty in words. This is done by adding a new output format that allows the model to provide a probability distribution over possible responses, along with a confidence score. The goal of this approach is to make it easier for users to understand when an AI's response may not be entirely accurate or reliable. --- Why it matters: This matters because it can help improve the reliability and trustworthiness of AI systems in applications where accuracy is critical, such as decision-making or expert advice. Source: https://openai.com/index/teaching-models-to-express-their-uncertainty-in-words