A software toolkit the Veritas Toolkit has been updated to assist financial institutions in evaluating the responsible use of artificial intelligence (AI). Developed by a consortium led by the Monetary Authority of Singapore (MAS), the toolkit aims to guide organizations in the financial sector by providing assessment methodologies for accountability and transparency, in addition to its existing focus on fairness and ethics. The Veritas consortium comprises 31 industry players, including prominent organizations such as the Bank of China, Google Cloud, Microsoft, Goldman Sachs, Visa, and IBM.
Expanding assessment components
The second iteration of the Veritas Toolkit builds upon its initial release in February of the previous year. The updated version now encompasses assessment methodologies for fairness, ethics, accountability, and transparency, collectively called FEAT. By leveraging the toolkit, financial institutions can define the objectives of their AI and data analytics implementations, identify potential biases, and ensure compliance with these essential principles.
MAS highlights that the second release of the Veritas Toolkit introduces an improved assessment methodology for the fairness component of FEAT and includes review methodologies for the remaining three principles. This comprehensive update addresses the diverse facets of responsible AI adoption, promoting an ecosystem that values fairness, ethics, accountability, and transparency. Several banks within the consortium have already tested the toolkit, affirming its practicality and effectiveness.
Veritas Toolkit 2.0 Open-source and integration capabilities
The Veritas Toolkit is an open-source solution available on GitHub, offering financial institutions the flexibility to customize and integrate it into their existing IT systems. The toolkit can be enhanced with plugins, facilitating seamless integration and enabling organizations to tailor the assessment processes to their specific requirements.
The consortium developed new use cases to showcase the Veritas Toolkit’s practical application. One such example involved Swiss Reinsurance, which conducted a transparency assessment of its predictive AI-based underwriting function. Google also shared its experience of implementing the FEAT methodologies in fraud detection payment systems in India and mapping its AI principles and processes. These use cases provide valuable insights and implementation lessons for other financial institutions looking to adopt responsible AI practices.
Whitepaper insights
In addition to the toolkit, Veritas released a whitepaper that outlines the experiences and lessons shared by seven financial institutions, including Standard Chartered Bank and HSBC. The document sheds light on integrating the AI assessment methodology with their internal governance frameworks. It emphasizes the importance of a responsible AI framework that spans different geographies and highlights the need for a risk-based model to determine appropriate governance measures for various AI use cases. The whitepaper also covers responsible AI practices and the significance of training a new generation of AI professionals in the financial sector.
Sopnendu Mohanty, Chief Fintech Officer at MAS, stresses the necessity for financial institutions to establish robust frameworks for the responsible use of AI due to the rapid pace of AI advancements. According to Mohanty, the Veritas Toolkit version 2.0 empowers organizations to assess their AI use cases effectively for fairness, ethics, accountability, and transparency.. The toolkit promotes responsible AI practices and develops an AI ecosystem that prioritizes ethical considerations.
Addressing risks and OpenAI’s perspective
The Singapore government has identified six key risks associated with generative AI and proposed a framework to mitigate these issues. OpenAI CEO Sam Altman, during his recent visit to Singapore, emphasized the importance of developing generative AI with public consultation, ensuring human control to mitigate potential risks and harm associated with AI adoption. Altman also highlighted the challenges related to bias and data localization, advocating for training AI platforms on diverse datasets encompassing various cultures, languages, and values.
The Veritas Toolkit 2.0 represents a significant step forward in the responsible use of AI for financial institutions. With its comprehensive assessment methodologies covering fairness, ethics, accountability, and transparency, the toolkit enables organizations to evaluate their AI implementations and ensure adherence to vital principles. By leveraging this open-source solution and the insights shared in the accompanying whitepaper, financial institutions can navigate the evolving landscape of AI technology, fostering responsible AI practices and contributing to developing an ethical and transparent AI ecosystem.
