Governance Unplugged – Episode 7
“Beyond the Hype: Using AI with Integrity in Internal Audit”
Guest: Gavin Martin, seasoned internal audit executive and technology risk expert with leadership roles at UBS, Credit Suisse, and Lloyds Banking Group.
How can we embrace Artificial Intelligence in Internal Audit, without losing our professional skepticism?
In this episode, host Thomas is joined once again by Gavin Martin to explore one of the most pressing topics in today’s assurance landscape: the rise of AI. Together, they unpack how internal auditors can leverage AI responsibly, without falling into the trap of blind adoption or over-reliance.
Drawing on decades of experience and sharp observation, Gavin explains how to distinguish between efficiency and effectiveness when applying AI, and why focusing on outcomes is more critical than ever.
Key Topics Discussed:
• The real value of AI: boosting efficiency and audit effectiveness
• Why outcome thinking must come before AI adoption
• How unconscious biases and “authority bias” can mislead auditors
• What “AI drift” is, and how to spot it before it erodes trust
• Risks of over-relying on AI-generated audit evidence and reports
• How internal audit can audit AI-enabled business processes
• What it takes to build responsible AI governance frameworks
• Deepfakes, model risk, and the return of old-school audit techniques
• Human roles in an AI-driven audit world: recalibrating expectations
Takeaway:
Adopting AI is inevitable, but it must be done with clarity, curiosity, and integrity. Don’t just follow the trend. Define the outcome, align the tools, and empower the humans behind the process.
Gavin’s Advice to Auditors:
“Focus on the outcome. Know whether you're targeting efficiency, effectiveness, or both. Then ask the right questions, of the AI, and of yourself.”
Don’t forget to follow or subscribe to Governance Unplugged on your favorite podcast platform and connect with host Thomas Michel on LinkedIn. Have a guest or topic in mind? Visit keystone-governance.ch to get in touch.