AI-Driven ESG Analytics for Sustainable Investment in U.S. Small Businesses: Integrating LLMs and Causal Modeling for Policy-Enhanced Resilience

Abstract

This study examines how artificial intelligence, particularly large language models (LLMs) and causal inference frameworks, can enhance ESG-oriented investment decision-making for U.S. small businesses. As sustainability criteria increasingly influence capital allocation, investors require analytical systems capable of processing fragmented disclosures, unstructured narratives, regulatory documents, and heterogeneous financial indicators. The research develops an integrated AI-driven ESG analytics framework that leverages automated text understanding, probabilistic causal modeling, and resilience forecasting to identify sustainability patterns while remaining policy-neutral. Empirical evaluation using synthetic and publicly available datasets indicates that LLM-enhanced ESG scoring improves signal extraction from incomplete disclosures, while causal models clarify the directional impact of environmental, social, and governance factors on financial stability. The combined system demonstrates strong potential for supporting investors, policymakers, and financial institutions in assessing long-term resilience among small enterprises. The findings highlight AI's transformative role in sustainability analytics and provide pathways for future refinement through regulatory harmonization, domain-specific model alignment, and expanded cross-sector datasets.