Defining and evaluating political bias in LLMs (openai.com)

Describes methods for testing and evaluating political bias in large language models like ChatGPT to improve objectivity and reduce partisan outputs.