News

OpenAI explains why language models ‘hallucinate’; evaluation incentives reward guessing over uncertainty

OpenAI finds a key problem in how large language models work. These models often give wrong information confidently. The ...