EP 560: Inside Multi-Agentic AI: 3 Critical Risks and How to Navigate Them

In this episode of the Everyday AI Show, Jordan Wilson interviews Sarah Bird, Chief Product Officer of Responsible AI at Microsoft, about the evolving landscape of responsible AI in the age of agentic and multi-agentic systems. They discuss how the increasing autonomy of AI agents necessitates new approaches to ethics, governance, and risk management. Sarah explains how Microsoft is adapting its tools and technologies to manage these new entities, secure them, and address potential risks like agents going off-task or leaking sensitive data. The conversation covers the importance of testing AI systems, monitoring agent behavior, and preparing the workforce for collaboration with AI agents, emphasizing the shift from human-in-the-loop to human-on-the-loop models. Sarah also highlights the need for business leaders to be intentional, invest in testing, and understand the different categories of risks associated with agentic AI, including malfunctions, misuse, and systemic risks.