人工智慧金融與金融歧視之研究—以信用貸款為中心

Abstract

隨著人工智慧技術的發展，其於各行業中的應用日益普及，金融機構也不例外地在金融服務中開始運用人工智慧技術，藉此提升效率以提供更完善的金融服務。但於此同時，人工智慧技術的運用也帶來了公平性的隱憂。 本文以金融機構運用人工智慧技術於信貸服務中引發的歧視問題為討論主軸，從目前實務上發生的案例，發現人工智慧信貸不同於傳統信貸的歧視風險態樣，進而梳理人工智慧信貸歧視的風險形成與原因，並指出我國現行金融反歧視法存在規範空白的問題。本文進而透過比較法研究討論實務和學說如何在現行金融歧視法規範的架構下解決人工智慧信貸歧視問題，並以美國公平信貸相關規範及學者研究文獻為主要參考對象，觀察傳統公平信貸法制實踐於人工智慧信貸時產生的公平性問題與解釋挑戰。 透過比較法的研究，本文歸納我國法制針對人工智慧信貸應用可考慮的監理路徑，並以《金融業運用人工智慧（AI）指引》及《金融機構運用人工智慧技術作業規範》的規範內容為主，提出我國金融監理面對人工智慧信貸歧視問題時可考慮的具體規範建議。

With the development of Artificial Intelligence (AI) technology, its application in various industries is becoming more and more popular. Financial institutions are no exception, which have started to utilize AI technology in financial services to enhance efficiency and provide better financial services. However, at the same time, the use of AI technology in financial services also brings about the hidden problem of fairness. This thesis focuses on the use of AI technology in credit services by financial institutions. It identifies the different patterns of discrimination of AI credit services from that of traditional credit services from the cases occurring in practice, explores the causes of AI credit discrimination, and specifies the gaps in the existing financial anti-discrimination laws in Taiwan. Furthermore, the thesis adopts the comparative legal study to discuss how the practice and theory can solve the problem of AI credit discrimination under the existing financial anti-discrimination legal framework in other countries. It mainly focuses on the legal reforms in the U.S. related to fair credit and scholars' literatures to observe the interpretation challenges related to the fairness problem when applying the traditional fair credit laws to AI credit services. This thesis finally discusses how the current legal system can be applied in the context Taiwan's law, specifically the Guidelines on the Use of Artificial Intelligence (AI) in the Financial Sector and the Code of Practice on the Use of Artificial Intelligence by Financial Institutions. It proposes specific regulatory suggestions to address the problem of discrimination in the use of AI in credit services.