Causality Elicitation from Large Language Models

Takashi Kameyama, Masahiro Kato, Yasuko Hio +2

Large language models (LLMs) are trained on enormous amounts of data and encode knowledge in their parameters. We propose a pipeline to elicit causal relationships from LLMs. Speci…