We assess the ability of large language models (LLMs) to answer causal questions by analyzing their strengths and weaknesses against three types of causal question. We believe that current LLMs can answer causal questions with existing causal knowledge as combined domain experts. However, they are not yet able to provide satisfactory answers for discovering new knowledge or for high-stakes decision-making tasks with high precision. We discuss possible future directions and opportunities, such as enabling explicit and implicit causal modules as well as deep causal-aware LLMs. These will not only enable LLMs to answer many different types of causal questions for greater impact but also enable LLMs to be more trustworthy and efficient in general.

我们分析了大型语言模型（LLMs) 回答因果问题的能力以及它们在三种类型的因果问题中的优势和劣势。我们认为，目前的LLMs可以结合现有的因果知识回答因果问题，但它们还不能满足于发现新的知识或高精度的决策任务。我们探讨了可能的未来方向和机会，例如启用显式和隐式因果模块以及深度因果感知LLMs。这些不仅能够使LLMs回答更多不同类型的因果问题，也能使LLMs在一般情况下更加值得信赖和高效。

利用大型语言模型理解因果关系：可行性和机遇