AI-Driven Evaluation of Environmental, Social, and Governance Disclosures

In recent years, environmental, social, and governance (ESG) reporting has emerged as a crucial metric for measuring and reporting progress toward the targets set out by the United Nations’ Sustainable Development Goals. Investors, stakeholders, and regulators increasingly demand accurate ESG evaluations to inform investment decisions, risk management, and compliance strategies. However, existing ESG scoring methodologies suffer from several limitations, including subjectivity, inconsistency, and high costs. To address these challenges, this study proposes an innovative AI-powered tool that utilizes unsupervised machine learning and NLP techniques to comprehensively score companies’ ESG metrics with accuracy, transparency, and efficiency. The study focuses on developing a prototype software that extracts data from various sources, including reports, websites, and social media. The aim of the study is to apply the Turing Test to determine whether a human can achieve the same ESG score as the AI tool. The paper commences with a literature review on ESG reporting and AI-powered tools that assess ESG disclosures. Thereafter, the design science research is applied in the development of the AI tool. Using the Turing Test, an experiment is conducted to evaluate whether the results obtained by the AI tool are similar to those of humans. The results of the Turing Test showed that humans and AI achieved similar scores in evaluating ESG disclosures. The design and development of the AI tool contribute to the development of more effective ESG evaluation methodologies, promoting sustainable investment practices, responsible business conduct, and better decision-making among stakeholders.