Tenile Collins, Chief Data Officer, and Lisa Keeling, General Manager - Telecommunications Modernisation, Automation and Architecture at Services Australiation in Brazil
Abstract
With growing social needs and rising uncertainty, early identification of people in vulnerable circumstances is ever-important. Yet vulnerability is often difficult to detect across high volumes of citizen interactions, especially when hidden is nuanced expressions, misspellings or indirect signals. AI offers a new avenue to identify those at risk and enable faster, more targeted support. Services Australia has introduced the “Feedback Escalation” model, combining machine learning, large language models and human oversight to identify vulnerability in customer complaints. Launched in March 2026, the system can detect vulnerability even when expressed indirectly or through nuanced expressions, and has undergone rigorous legal, privacy, cybersecurity and ethical assurance. Early results suggest earlier identification of vulnerable customers, faster access to support, improved allocation of complex cases, and reduced psychosocial risks for frontline staff.
Join Tenile Collins, Chief Data Officer, and Lisa Keeling, General Manager - Telecommunications Modernisation, Automation and Architecture at Services Australia, for this fascinating case study presentation, followed by a live Q&A moderated by Monika Queisser, Head of Social Policy at the OECD.