Causal MAS: A Survey of Large Language Model Architectures for Discovery and Effect Estimation

2025-08-31

AI Summary

Large language models (LLMs) suffer from hallucination, spurious correlation reliance, and poor domain adaptability in complex causal reasoning, discovery, and effect estimation. To address these challenges, we propose a multi-agent collaborative paradigm specifically designed for causal inference, integrating adversarial debate, interactive simulation environment grounding, explicit causal modeling, and iterative refinement. We establish the first taxonomy for causal multi-agent LMs—unifying architectural design principles, evaluation benchmarks, and application scenarios—and delineate concrete implementation pathways and core challenges in scientific discovery and healthcare. Empirical results demonstrate that our framework substantially improves accuracy, robustness, and interpretability across causal tasks, offering a novel methodological foundation for trustworthy causal AI.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Multiagent Systems: Adversarial Agents

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Large language models for search
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Large Language Models (LLMs) have demonstrated remarkable capabilities in various reasoning and generation tasks. However, their proficiency in complex causal reasoning, discovery, and estimation remains an area of active development, often hindered by issues like hallucination, reliance on spurious correlations, and difficulties in handling nuanced, domain-specific, or personalized causal relationships. Multi-agent systems, leveraging the collaborative or specialized abilities of multiple LLM-based agents, are emerging as a powerful paradigm to address these limitations. This review paper explores the burgeoning field of causal multi-agent LLMs. We examine how these systems are designed to tackle different facets of causality, including causal reasoning and counterfactual analysis, causal discovery from data, and the estimation of causal effects. We delve into the diverse architectural patterns and interaction protocols employed, from pipeline-based processing and debate frameworks to simulation environments and iterative refinement loops. Furthermore, we discuss the evaluation methodologies, benchmarks, and diverse application domains where causal multi-agent LLMs are making an impact, including scientific discovery, healthcare, fact-checking, and personalized systems. Finally, we highlight the persistent challenges, open research questions, and promising future directions in this synergistic field, aiming to provide a comprehensive overview of its current state and potential trajectory.

Problem

• Addressing LLMs' limitations in complex causal reasoning and discovery
• Overcoming hallucination and spurious correlations in causal relationships
• Developing multi-agent systems for nuanced domain-specific causal estimation

Innovation

• Multi-agent systems for causal reasoning
• Diverse architectural patterns and protocols
• Evaluation methodologies across application domains