Bias in, bias out: Lessons from the AI training ground

20 Aug 2026
WAFA | Meeting Room 210
Across Australia, communities increasingly expect timely, tailored and relevant warnings, particularly following major events such as the 2022 Victorian floods. In response, our consequence-based, intelligence-driven warnings project is transforming how flood and storm warnings are designed and delivered. Over a three-year period, we are combining literature, parliamentary inquiry findings and lived-experience research with community workshops to better understand what people need - and what they expect - from emergency services.

However, improved warnings present a critical tension: how do we meet rising community expectations without unintentionally reinforcing reliance on emergency services? How do we strengthen resilience and shared responsibility while providing clearer, more localised and more actionable information?

This presentation explores how we are navigating that balance. It examines how consequence-based messaging can increase clarity about impacts while reinforcing individual and community agency. It outlines how engagement processes are elevating diverse community voices and are shaping more inclusive communication approaches. It also reflects on lessons from recent flood events about trust, credibility and the role of transparency in public communication.

By sharing practical insights and challenges encountered, this session will offer delegates actionable strategies to enhance warnings while embedding resilience-building principles. Ultimately, leading together means delivering information with integrity - ensuring communities are informed, empowered and active partners in disaster preparedness and recovery.
 
Speakers
Sara Mills-Kerr
Sara Mills-Kerr, Firefighter/ PhD candidate, FRNSW/ University of Newcastle