Artificial intelligence in ESG investing: A scoring model for accuracy and accountability

College of Science, North Carolina State University, Raleigh, North Carolina, the United State

^* Corresponding author: apyulie@gmail.com

Abstract

This article meticulously examines the utilization of big data and artificial intelligence (AI) to tackle the significant challenges encountered in ESG (Environmental, Social, and Governance) investments. These challenges primarily include the inconsistent ESG ratings across different rating agencies and the lack of transparency in AI models, which can hinder informed decision-making. The study meticulously constructs an AI-based ESG scoring model by integrating advanced techniques such as natural language processing (NLP), topic modeling, and machine learning. It also proposes a comprehensive explainable AI framework to enhance investment confidence and provide clear insights into the decision-making process. According to the research reports, these techniques significantly boost the effectiveness and granularity of ESG analysis, allowing for more precise and reliable assessments. However, the study acknowledges persistent challenges such as algorithmic bias, data heterogeneity, style issues, and conceptual problems that need to be addressed. It provides some thoughtful recommendations for future research to tackle these issues effectively. This review aims to bolster investor and business confidence and improve the overall performance of ESG investing by promoting more transparent and consistent practices.