Preprint: Human-AI interaction in safety-critical network infrastructures: – extended abstract

Abstract

Artificial Intelligence (AI) has substantial potential to support decision-making in safety-critical network infrastructures. This extended abstract summarizes shared decision-making challenges across power grids, railway networks, and air traffic management and outlines a psychologically grounded perspective on complementary human-AI collaboration. Particular attention is given to transparency as a means of supporting operators’ situation awareness, mental models, learning, trust, and meaningful involvement in decision-making. We argue that AI-based systems in critical infrastructures should not focus on automation alone, but should support human operators as part of a joint cognitive system that leverages the complementary strengths of humans and AI.