Human-Ai Hybrids in Safety-Critical Systems: Concept, Definitions and Perspectives from Air Traffic Management

Abstract

The recent advancements in Artificial Intelligence (AI) have paved the way for Human-AI Hybrid (HAH) systems, which integrate human and AI capabilities to augment human ingenuity rather than replace it. However, the application of HAH in Safety-Critical Systems (SCS), such as Air Traffic Management (ATM), remains limited due to the high stakes involved and the challenges presented by AI's unpredictable behavior and limitations in complex reasoning tasks. This paper provides an extensive review of the emerging domain of HAH in SCS, defining key concepts and examining the fundamental pillars of effective HAH in SCS, including collaboration, adaptation, and trust. By synthesizing interdisciplinary research, this review explores the interaction between humans and AI, identifies obstacles, and recommends strategies for developing effective HAH systems in SCS. Furthermore, by examining real-world ATM applications, this study bridges the gap between theoretical recommendations and practical implementation, offering valuable insights for future efforts in similar contexts.

Keywords: Human-AI Hybrid, Safety-Critical Systems, Collaboration, Adaptation, Trust, Air traffic management