… defined causal graphs required for formal causal reasoning. To over… and unified definition
of causality involving interactions … of 14,094 causal questions across all four levels of causality: …

… This paper explores the causal reasoning capabilities of Large Language Models (LLMs) …
focusing on tasks like causal DAG generation, counterfactual reasoning, and token causality. It …

… ), it addresses the limitations of existing causal datasets and offers a … causal reasoning
abilities of language models. The findings in this paper suggest that models with stronger causal …

… 2 Causal Inference and Large Language Models … to both large language models (LLMs)
and causal inference, laying … natural language processing by enabling sophisticated language …

… perform causal reasoning is widely considered a core feature of intelligence. In this work,
we investigate whether large language models (LLMs) can coherently reason about causality. …

… language models for causal DAG generation, counterfactual reasoning, and token causality
demonstrates that they bring significant new capabilities across a wide range of causal tasks…

… 4 Evaluations of LLMs in Causal Tasks This section summarizes recent evaluation results
of LLMs in causal tasks. We mainly focus on causal discovery and causal effect estimation, …

… 3 Towards Causal Reasoning in Large Language Models We separate the roles of language
models in CR into two categories. First, LLMs can serve as causal reasoning engines, …

… We present the CARE-CA framework as a significant advancement in enhancing the causal
reasoning capabilities of large language models (LLMs). By integrating explicit knowledge …

… ’ causal capabilities, we introduce a novel evaluation method for assessing their causal …
This section presents the evaluation results of the selected LLMs on three causal reasoning …