Lunch and Learn: Scalable oversight of AI
This session* presents two complementary lines of work addressing challenges. First, a system-level responsible AI engineering approach that bridges governance and engineering, including methods for distinguishing intrinsic risk from design risk, a question bank for structured risk assessment, and multi-layered guardrail frameworks. Second, verifiability-first AI engineering, a paradigm where problems are deliberately decomposed so that verification can be scalable and stackable. Together, these contributions offer a practical path toward scalable oversight of AI systems that supports both public accountability and responsible adoption of AI.
*Please note this event will be recorded. You can view this video and other MasterCraft Series events here.
Presenter
Dr. Qinghua Lu is the Acting Research Director of Software and Computational Systems (SCS) Research Program at CSIRO and a Senior Principal Research Scientist specialising in AI engineering and responsible AI.
Qinghua is the only member from Australia among the 40 experts appointed to the UN’s Independent International Scientific Panel on AI, the first global scientific body dedicated entirely to AI. She has led/co-led major national AI initiatives, including Australia’s Voluntary AI Safety Standards and Department of Industry’s General-Purpose AI Risk Assessment Methodology. Internationally, she represents Australia in key international AI safety initiatives, including the Frontier Model Forum. She was the winner of Asia-Pacific Women in AI Trailblazer Award in 2023.
She has published over 200 papers in premier international journals/conferences. She is the lead author of the world’s first practitioner-focused Responsible AI book, Responsible AI: Best Practices for Creating Trustworthy AI Systems, which reached #3 Amazon’s AI best-seller, and the co-author of Engineering AI Systems: Architecture and DevOps Essentials. Her Responsible AI Pattern Catalogue has been adopted by Australia’s National AI Centre Digital Pathways and the National Framework for the Assurance of AI in Government and is widely used by industry to improve AI products or assess AI frameworks.
Facilitator
Anthony Dusan, APS Academy
Participant benefits
Learn practical approaches to bridge the gap between AI governance frameworks and engineering practice.
Methods for distinguishing intrinsic risk from design risk and a structured question bank for risk assessment, moving beyond high-level principles to actionable tools.
Gain insight into verifiability-first AI engineering, a new paradigm for designing AI systems where verification is scalable and stackable, offering a practical path to meaningful oversight without creating bottlenecks that slow down responsible AI adoption.
Suitable for
All Staff
Category and User level
This learning experience aligns with the Digital Profession at the Foundation level.
Price
Free of charge.