Cognitive Core Banking: A Data-Engineered, AI-Infused Architecture for Proactive Risk Compliance Management

Abstract

The long list of AI governance challenges motivates the need for new solution approaches. A portion of these issues are general to deploying AI models. An exploration of the broader concerns attempting to address the high-level risks of AI use in financial services is presented. There is however a perceived gap between those objectives, and the practical applications supporting the mass adoptions of AI systems in the sector. The focus is on the latter. It is acknowledged that a vast amount of work remains to deliver the promised economic potential for society, first and foremost the widespread understandability and trustable AI systems. The financial services industry is undergoing a profound transformation from static compliance to continuous adjustment of short-term risks. Such transformation requires a decision-making system that can well respond to the proactively evolving operational environment in real time.

Such awareness has pointed out a long list of risks induced or amplified by AI systems. It is agreed that improper use can result in severe harm and entanglement. Critics have broadly categorized the hazards falling into bias, security, accountability, and oversight. The financial services are a significant application area of AI systems. This broad “financial services” comprises a vast ecosystem that includes different business activities involving personal, institutional or governmental finance. Several of the breakthroughs in machine learning and artificial intelligence have been leveraged by the finance sector, which has long been a source of fundamental research questions, particularly in the areas of prediction, time series, and optimization. Today most of the leading global financial institutions have adopted or are exploring the adoption of AI systems across their business activities, for B2B or B2C. The variety and complexities of AI systems used in the financial services, however, have also revealed a range of new challenges.