Environmental, social, and governance sustainability: an AI-centric approach driving data standardization and automation
環境・社会・ガバナンスのサステナビリティ：AI中心のアプローチによるデータ標準化と自動化の推進 (AI 翻訳)
A. Telukdarie, M. Nyathi, R. J. Fabchi
🤖 gxceed AI 要約
日本語
本論文は、ESG報告の不整合やグリーンウォッシュ問題に対処するため、NLPとデータ管理システムを統合したAI中心の枠組みを提案。南アフリカとインドの440社の報告書を用いてDistilRoBERTaモデルを微調整し、環境・ガバナンス指標を高精度で分類した。社会的指標の精度は低いものの、AIによるESG開示の標準化と比較可能性向上の可能性を示す。
English
This paper proposes an AI-centric framework using NLP and centralized data management to standardize ESG reporting and combat greenwashing. A DistilRoBERTa model fine-tuned on sustainability reports from 440 firms in South Africa and India achieved 99.1% accuracy for environmental indicators and 99.3% for governance, but only 63.6% for social indicators, highlighting challenges in qualitative disclosures. The approach demonstrates potential for real-time, standardized ESG insights to support regulatory oversight and global data governance.
Unofficial AI-generated summary based on the public title and abstract. Not an official translation.
📝 gxceed 編集解説 — Why this matters
日本のGX文脈において
日本企業はESG報告の標準化とSSBJへの対応が急務だが、本論文のAIによる分類手法は、開示データの自動チェックや比較可能性向上に示唆を与える。特に、定性情報の多い社会指標の精度向上が今後の課題として浮き彫りになった点は、日本の統合報告書作成にも参考となる。
In the global GX context
This paper directly addresses the global challenge of ESG data comparability and greenwashing, which are central to ISSB, CSRD, and SEC climate disclosure rules. Its AI-driven approach, tested on emerging market data, offers a scalable solution for automating ESG indicator extraction and reducing human error, supporting the move toward standardized, machine-readable disclosures worldwide.
👥 読者別の含意
🔬研究者:Shows how fine-tuned NLP models can classify ESG indicators from corporate reports, with notable accuracy gaps for social metrics that warrant further investigation.
🏢実務担当者:Provides a proof-of-concept for automating ESG data extraction and standardization, which can reduce manual effort and improve consistency in sustainability reporting.
🏛政策担当者:Illustrates how AI can enhance regulatory oversight of ESG disclosures and support the development of standardized reporting frameworks.
📄 Abstract（原文）
Inconsistent Environmental, Social, and Governance (ESG) reporting and widespread greenwashing have undermined transparency, trust, and comparability in global sustainability assessments. This study proposes an AI-centric framework to support automated analysis and improved standardization of ESG reporting by integrating Natural Language Processing (NLP) and centralized data management systems. Using sustainability reports from 440 companies listed on the Johannesburg Stock Exchange (JSE) and the National Stock Exchange of India (NSE), a DistilRoBERTa transformer model was fine-tuned to classify ESG indicators. The model achieved an accuracy of 99.1% for environmental indicators, 99.3% for governance, and 63.6% for social indicators, reflecting the qualitative and heterogeneous nature of social disclosures. These results demonstrate that AI can reduce subjectivity, increase comparability, and minimize human error in ESG disclosures. By enabling real-time, standardized ESG insights, this framework supports regulatory foresight, enhances global data governance, and advances policy-oriented sustainability planning.
🔗 Provenance — このレコードを発見したソース
- semanticscholar https://doi.org/10.3389/frsus.2026.1793155first seen 2026-05-15 18:45:33
🔔 こうした論文の新着を逃したくない方は キーワードアラート に登録（無料・3キーワードまで）。
gxceed は公開メタデータに基づく研究支援データセットです。要約・翻訳・解説は AI 支援で生成されています。 最終的な解釈・検証は利用者が原典資料に基づいて行うことを前提とします。