CASE STUDY
Evolving Financial Services
The Monetary Authority of Singapore and CDWT founded the Veritas industry consortium in order to propose innovative principles for Responsible AI.
Call For Change
In today’s financial services market, organizations increasingly depend on Artificial Intelligence (AI) to deliver superior customer service while maintaining lean operations and cheap costs.
The Monetary Body of Singapore (MAS), Singapore’s central bank and financial regulatory authority, has acknowledged the advantages that artificial intelligence (AI) offers to financial services organizations (FSIs). However, it was also cognizant of the possible influence of AI’s unforeseen repercussions on the sector.
These dangers may include AI algorithms mistakenly rejecting proportionately more persons of a specific sex, color, or religion for credit card applications, or residents of a certain neighborhood being charged higher insurance premiums when the claim rates do not warrant it. MAS was aware that when FSIs addressed these concerns, they would be confronted with challenging considerations about ethics, accountability, and transparency.
MAS’s mission is to assist FSIs in using AI safely and mitigating unwanted repercussions.
When Tech Meets Human Inventiveness
As one of the first financial regulators with a dedicated Responsible AI program, MAS enables FSIs to assess their AI and data analytics solutions based on the core values of justice, ethics, accountability, and transparency (FEAT).
MAS founded and directed Veritas, an industry consortium with over 25 members, to co-develop a methodological framework for operationalizing the FEAT principles.
A core team within the consortium Veritas created three important approaches to support the framework. Consistently, feedback was included from the larger collaboration. This assured that the final results would be solid, applicable, and applicable to FSIs of different sizes and levels of maturity. The core group:
Defined a new Ethics and Accountability Assessment Methodology, which provides a framework for defining ethical commitments, justice conceptions, and principles.
Extended and modified the Fairness Assessment Methodology to enable FSIs to describe the fairness goals of their systems, detect individual characteristics and bias, and design mitigation solutions.
Defined a Transparency Assessment Methodology to aid FSIs in determining whether and how much transparency is required to evaluate the predictions of machine learning models
Five use cases were developed for each methodology to demonstrate their applicability.
The CDWT programme team then established the overall framework into which these three techniques and the consolidated collection of FEAT checklist questions fit and connected them with a normal AI system development lifecycle, so allowing “FEAT by design.” The overall framework also provides critical basic practises, such as a risk-based approach to FEAT assessments, to assist businesses in scaling FEAT in a realistic manner.
A Valuable Difference
MAS is the first regulator to establish such a comprehensive framework for FEAT, and its advice enables FSIs to go from principles to practice.
MAS and CDWT continue to collaborate throughout the ecosystem of financial services to:
- Provide open-source toolkits to assist FSIs in defining a common FEAT assessment set
- Provide materials and webinars to build a group of individuals with knowledge and experience in executing MAS’s recommendations.
- urge other technology firms to develop AI solutions in accordance with the FEAT principles
- Consult with FSIs that are using FEAT approaches and apply the lessons learned to strengthen the methodology.
Through the Veritas consortium, MAS and CDWT are influencing the financial services sector as a whole. They are advancing the use of AI by FSIs that is fair, ethical, responsible, and transparent. And they are assisting FSIs in ethically gaining value from AI and creating a fairer future for the benefit of billions of customers globally.