人工知能・説明責任・サステナビリティ開示：サウジアラビアのイスラム金融機関に関する社会技術的分析

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

🔬研究者:AI支援開示分析と手動評価の相補性、およびESG–シャリーア統合枠組みの方法論的貢献を学べる。

🏢実務担当者:AIによる開示セマンティック分析を自社のESG・統合報告書品質評価に応用する際の限界と有効性を理解できる。

🏛政策担当者:AI支援開示評価の標準化や、暗黙的・文脈依存的開示の評価における手法選択の留意点を検討する材料になる。

This study examines sustainability disclosure practices in Saudi Islamic financial institutions through a comparative assessment of manual content analysis, artificial intelligence (AI)-assisted semantic evaluation. Drawing upon a hybrid environmental, social, governance (ESG)–Shariah framework, the study evaluates disclosure quality across five dimensions: environmental, social, governance, Shariah compliance, Maqasid al-Shariah. The analysis covers the complete population of fully Shariah-compliant commercial banks operating in Saudi Arabia (four banks) over the period 2022–2024. A structured disclosure index is employed for manual assessment, while AI-assisted techniques are used to analyze semantic intensity, thematic emphasis, narrative sentiment within sustainability reports. The findings reveal that sustainability disclosure is governance-driven, socially oriented, whereas environmental, explicitly articulated Maqasid-related disclosures remain comparatively less developed. Manual, AI-assisted approaches identify broadly similar disclosure patterns while exhibiting distinct interpretive sensitivities. The overall mean absolute difference (MAD) of 7.34 percentage points indicates broad cross-method convergence without implying methodological interchangeability, while the larger divergence observed for Maqasid Impact (MAD = 13.80 percentage points) highlights the challenge of evaluating implicit, ethically embedded, context-dependent disclosure through general-purpose semantic models. In addition, sensitivity analysis under alternative weighting scenarios confirms the robustness of the assessment framework, with the overall ranking of banks remaining stable. The study contributes methodologically by demonstrating the complementary value of integrating structured manual evaluation with AI-assisted semantic analysis rather than treating the two approaches as substitutes. Theoretically, it advances the application of an integrated ESG–Maqasid perspective within Islamic finance. Practically, the findings provide insights for regulators, standard setters, Islamic financial institutions seeking to strengthen sustainability disclosure quality, improve transparency, enhance ESG–Maqasid reporting practices

• semanticscholar https://link.springer.com/content/pdf/10.1186/s43093-026-00986-4.pdffirst seen 2026-09-12 05:25:42 · last seen 2026-09-21 05:01:20