Extended Data Fig. 1: Example prompts for multiple choice and short-answer questions.
From: What large language models know and what people think they know
Example prompt to elicit model confidence for a multiple-choice question (top) and short-answer question (bottom). For the multiple-choice question, the prompt elicits the answer and model confidence across answer options. Note that for the short-answer question, the LLM answer (dove) was elicited through another prompt.
