Model Risk Management Evolves Amid AI and Analytical Tool Explosion

vTingting She provides a compelling overview of how traditional model risk management frameworks are being tested by the rise of AI and advanced analytics.

While the foundational definition of a "model" remains unchanged, the tools and technologies now in use—ranging from simple spreadsheets to generative AI—blur the lines between what qualifies as a model versus a non-model, creating challenges in governance and compliance. The rapid expansion of analytical tools and AI applications requires firms to rethink their inventories and risk frameworks.

She outlines a dual-path forward: a short-term bottom-up approach that assesses each AI application individually to determine the appropriate governance framework, and a long-term top-down strategy that calls for a flexible umbrella framework. This would encompass all analytical tools—models, EUCs, and AI-driven systems alike—without being constrained by outdated regulatory boundaries.

The session emphasizes the importance of balancing risk control with practicality, especially when traditional documentation and validation processes may not suit the scale and complexity of modern AI systems.