Design Thinking Framework for Integration of Transparency Measures in Time-Critical Decision Support

Abstract

The integration of artificial intelligence transparency in time-critical decision support is complex and requires consideration of the impact on human-machine teaming. The relationships between transparency, trust, workload, and situational awareness are key to understanding this impact on performance. We detail the development of a novel design framework for transparency integration in Decision Support Systems. We selected the design thinking approach as the baseline for our framework as this focuses on developing empathy with users and rapid design iteration. We adapted this framework by introducing the concept of empathy for both human and machine agents. In this situation, “empathy” is providing a deep understanding of the model, its purpose and the underlying data for AI. We developed a structured problem definition focused on understanding the relationships between constructs and established solution themes to guide the designer. We demonstrate this transparency integration framework on a Transfer of Care Decision Support System.