In this inaugural episode of The Savvy CIO, host Bradd Busick speaks with Dr. Radha Plumb, IBM’s VP of AI-first transformation and former Pentagon chief digital and AI officer, about making real-world AI deployments audit ready. She breaks down the tough questions, like: How does your LLM differ from the legacy systems your auditor is used to looking at? Which amorphous risks do you need to be on the lookout for if you want to keep up the pace of progress? Can ensuring your systems are doing what you want them to do actually enable speed in your AI governance?
They also discuss end-to-end compliance hurdles: role-based access control and data access decisions; the need for an orchestration layer to route data between LLM and deterministic methods; and why many failures occur at this intersection between technology and process.
Using the IBM model, where they’re “taste testing their own cooking”, Dr. Plumb emphasizes model transparency, the importance of bounded data for regulated decisions, why early CISO involvement is critical in designing for security, and why documenting enterprise workflows allows for not just a faster AI deployment - but one that can actually withstand an audit.
Highlights:
02:53 Meet Dr Radha Plumb
06:17 Why are auditors & LLMs fundamentally at odds?
14:07 What’s the difference between governing the data versus governing the LLM itself?
18:40 Is it possible to have speed and security in AI?
Links:
Park Place Technologies: https://www.parkplacetechnologies.com/
The Savvy CIO podcast page: https://www.parkplacetechnologies.com/the-savvy-cio/
Bradd Busick: https://www.linkedin.com/in/bbusick/
Facebook: https://www.facebook.com/ParkPlaceTech
LinkedIn: https://www.linkedin.com/company/park-place-technologies/
X: https://www.linkedin.com/company/park-place-technologies/
YouTube: https://www.youtube.com/parkplacetechnologies
Instagram: https://www.instagram.com/parkplacetechnologies/