ARCADIA: Scalable Causal Discovery for Corporate Bankruptcy Analysis Using Agentic AI

TL;DR

ARCADIA is a novel agentic AI framework that combines large-language-model reasoning with statistical diagnostics to construct accurate, interpretable causal models for corporate bankruptcy analysis, outperforming existing methods.

Contribution

The paper introduces ARCADIA, a scalable agentic AI approach that iteratively refines causal structures using LLM-guided prompting and causal validity feedback, advancing autonomous scientific modeling.

Findings

ARCADIA produces more reliable causal graphs than existing algorithms.

The framework offers a fully explainable and intervention-ready causal modeling pipeline.

Experiments demonstrate superior performance on corporate bankruptcy data.

Abstract

This paper introduces ARCADIA, an agentic AI framework for causal discovery that integrates large-language-model reasoning with statistical diagnostics to construct valid, temporally coherent causal structures. Unlike traditional algorithms, ARCADIA iteratively refines candidate DAGs through constraint-guided prompting and causal-validity feedback, leading to stable and interpretable models for real-world high-stakes domains. Experiments on corporate bankruptcy data show that ARCADIA produces more reliable causal graphs than NOTEARS, GOLEM, and DirectLiNGAM while offering a fully explainable, intervention-ready pipeline. The framework advances AI by demonstrating how agentic LLMs can participate in autonomous scientific modeling and structured causal inference.