Navigating Risk in the Age of Intelligent Applications

Rodanthy Tzani's presentation emphasizes the increasing importance of risk management frameworks for AI applications, including both traditional AI models and GenAI use cases. She underscores that the risks associated with AI stem from the complexity of models, predictability, unintended biases, and potential financial and legal repercussions.

Tzani highlights the evolving regulatory landscape, particularly within the insurance industry, with examples from NAIC, Colorado State, and DFS. These regulations emphasize governing principles, bias testing, and explicit oversight. To address these challenges, she advocates for enhancing risk management frameworks and implementing effective controls, which include publishing guiding AI principles and establishing governance building blocks.

To operationalize AI governance, Tzani suggests developing a governance framework with a clear committee structure, well-defined roles and responsibilities, and an AI inventory with risk assessments. She emphasizes the importance of prioritizing AI applications based on risk levels and introduces the concept of a bias testing framework, particularly relevant for insurance companies.