Tkachenko Nataliya
Smith School of Enterprise and the Environment, University of Oxford, Oxford, United Kingdom.
UK Centre for Greening Finance and Investment, University of Oxford, Oxford, United Kingdom.
Front Artif Intell. 2024 Jan 9;6:1168749. doi: 10.3389/frai.2023.1168749. eCollection 2023.
This paper delves into the intricacies of synthetic data, emphasizing its growing significance in the realm of finance and more notably, sustainable finance. Synthetic data, artificially generated to simulate real-world data, is being recognized for its potential to address risk management, regulatory compliance, and the innovation of financial products. Especially in sustainable finance, synthetic data offers insights into modeling environmental uncertainties, assessing volatile social and governance scenarios, enhancing data availability, and protecting data confidentiality. This critical review attempts first ever classification of synthetic data production methods, when applied to sustainable finance data gaps, elucidates the methodologies behind its creation, and examines its assurance and controls. Further, it identifies the unique data needs of green finance going forward and breaks down potential risks tied to synthetic data utilization, including challenges from generative AI, input quality, and critical ethical considerations like bias and discrimination.
本文深入探讨了合成数据的复杂性，强调了其在金融领域，尤其是可持续金融领域日益增长的重要性。合成数据是人为生成以模拟现实世界数据的，因其在解决风险管理、合规监管以及金融产品创新方面的潜力而受到认可。特别是在可持续金融中，合成数据有助于对环境不确定性进行建模、评估动荡的社会和治理情景、提高数据可用性以及保护数据机密性。这篇批判性综述首次尝试对应用于可持续金融数据缺口的合成数据生产方法进行分类，阐明其创建背后的方法，并审视其保障措施和控制手段。此外，它还确定了未来绿色金融独特的数据需求，并剖析了与合成数据利用相关的潜在风险，包括来自生成式人工智能、输入质量以及诸如偏差和歧视等关键伦理考量带来的挑战。
