The research paper presents a new methodology for causal inference that integrates large language models (LLM) into statistical causal discovery (SCD). This is achieved through statistical causal prompting (SCP) for LLMs and prior knowledge augmentation for SCD. The study shows that GPT-4 can improve the output of the LLM-KBCI and the SCD result with prior knowledge from LLM-KBCI. It also indicates that SCD results can be further enhanced if GPT-4 undergoes SCP. The approach can address challenges like dataset biases and limitations, demonstrating the potential of LLMs to enhance data-driven causal inference across various scientific domains.
Publication date: 5 Feb 2024
Project Page: Not provided
Paper: https://arxiv.org/pdf/2402.01454
