Seeing Isn't Believing: Trust, Transparency, and Judgment in AI Safety
Tracks
Room 1
| Tuesday, December 1, 2026 |
| 11:05 AM - 12:25 PM |
| Plenary Room - Grand Central |
Overview
Workshop - Dr Nicholas Duck
Speaker
Dr Nicholas Duck
Seeing Isn't Believing: Trust, Transparency, and Judgment in AI Safety
Abstract
Speaker: Dr Nicholas Duck, Registered Organisational Psychologist and Director, Opposite
Aim: AI-assisted safety systems are increasingly influencing operational decisions in heavy vehicle environments, yet their effectiveness frequently depends on factors the technology alone cannot address. Drawing on experience from Drive S(AI)FE, a multi-phase research program funded through the National Heavy Vehicle Regulator's (NHVR) Heavy Vehicle Safety Initiative and supported by the Australian Government, the workshop examines what happens when the fit between AI output and human judgment breaks down, and what good design looks like when it does not. The workshop applies human factors and ergonomics principles to designing the perfect fit between intelligent systems and the humans who operate them.
Workshop Plan: Participants begin by experiencing the problem directly. Using their own devices, they access a purpose-built interface presenting an AI fatigue monitoring alert for a specific driver in a specific operational context. They make an initial decision: act, request more information, or override. No framing is provided. The facilitator presents the same alert under three conditions: high transparency, low transparency, and ambiguity-inducing output. Participants respond to each version in real time, with responses displayed live showing how decisions shifted as transparency changed. A structured debrief examines trust calibration, explainability, and the conditions under which human judgment and AI output become aligned, misaligned, or break down in safety-critical moments. Participants then co-design trust calibration principles grounded in what they experienced rather than what they assumed. The format is adaptable to varying session lengths and audience sizes while retaining the core experiential and co-design components.
Outcomes for Participants: Participants leave with an experiential understanding of how transparency influences trust and decision-making in AI-supported safety systems, co-designed trust calibration principles applicable across safety-critical industries, and a framework for the procurement, implementation, or evaluation of AI-enabled operational technologies.
Aim: AI-assisted safety systems are increasingly influencing operational decisions in heavy vehicle environments, yet their effectiveness frequently depends on factors the technology alone cannot address. Drawing on experience from Drive S(AI)FE, a multi-phase research program funded through the National Heavy Vehicle Regulator's (NHVR) Heavy Vehicle Safety Initiative and supported by the Australian Government, the workshop examines what happens when the fit between AI output and human judgment breaks down, and what good design looks like when it does not. The workshop applies human factors and ergonomics principles to designing the perfect fit between intelligent systems and the humans who operate them.
Workshop Plan: Participants begin by experiencing the problem directly. Using their own devices, they access a purpose-built interface presenting an AI fatigue monitoring alert for a specific driver in a specific operational context. They make an initial decision: act, request more information, or override. No framing is provided. The facilitator presents the same alert under three conditions: high transparency, low transparency, and ambiguity-inducing output. Participants respond to each version in real time, with responses displayed live showing how decisions shifted as transparency changed. A structured debrief examines trust calibration, explainability, and the conditions under which human judgment and AI output become aligned, misaligned, or break down in safety-critical moments. Participants then co-design trust calibration principles grounded in what they experienced rather than what they assumed. The format is adaptable to varying session lengths and audience sizes while retaining the core experiential and co-design components.
Outcomes for Participants: Participants leave with an experiential understanding of how transparency influences trust and decision-making in AI-supported safety systems, co-designed trust calibration principles applicable across safety-critical industries, and a framework for the procurement, implementation, or evaluation of AI-enabled operational technologies.
Biography
Dr Nicholas Duck is a registered organisational psychologist and Director of Opposite, a consulting practice specialising in human-centred change across safety-critical industries. He currently leads Drive S(AI)FE, a multi-phase research program funded through the National Heavy Vehicle Regulator's (NHVR) Heavy Vehicle Safety Initiative and supported by the Australian Government. The program applies human-centred design principles to AI-assisted fatigue monitoring in heavy vehicle operations, with a specific focus on trust, transparency, and the conditions under which human operators and intelligent systems achieve, or fail to achieve, effective alignment.
With over fourteen years of experience spanning organisational psychology, human factors engineering, safety culture development, and process simplification, Nick has delivered specialist services across major Australian transport infrastructure projects including the Westgate Tunnel, Metro Tunnel, and Level Crossing Removal Project. His work across transport and rail has focused on decision-making, safety culture, and human-system integration under operational pressure. He has also worked with VicRoads, John Holland, and rail operators on operational decision-making and system usability.
His work sits at the intersection of human factors and organisational psychology, with a strong emphasis on how technology shapes attention, trust, and decision-making in real-world operational settings.
In this workshop, Nick translates the Drive S(AI)FE research into an immersive, evidence-informed experience. Participants interact directly with a purpose-built AI fatigue monitoring interface, making real-time decisions as transparency shifts from high to low to ambiguity-inducing output. He guides participants through structured debriefing on trust calibration, explainability, and decision bias, before facilitating small groups to co-design actionable trust principles grounded in lived experience rather than assumption.
His approach prioritises psychological safety and practical applicability, ensuring participants leave with tools and frameworks to evaluate and improve the alignment between intelligent systems and human judgment in safety-critical environments.
