On the Establishment of Trust: Rethinking Trust in the Age of Artificial Intelligence

Abstract

Trust has always been a fundamental element of organisational and social structures, enabling cooperation in situations where uncertainty or vulnerability cannot be fully eliminated. As societies and infrastructures have become increasingly interconnected, the foundations of trust have shifted from interpersonal relationships toward complex institutional and technological arrangements. This evolution has reached a new turning point: digital infrastructures, autonomous systems, and data‑driven decision‑making processes increasingly mediate interactions that once relied on human reasoning. Artificial intelligence (AI) is reshaping expectations of reliability, transparency, and accountability, prompting a reassessment of what it means to trust in environments where human and machine actions are intertwined. In this context, the notion of trustworthy AI has emerged as a key prerequisite for the responsible integration of AI systems into social and technical ecosystems. Core principles such as robustness, explainability, fairness, and security are widely regarded as essential for fostering a balanced relationship of trust. These requirements become especially significant in domains where AI supports or controls essential services. This is particularly true for critical infrastructures                       -energy grids, transportation systems, healthcare services, and communication networks-, where AI‑driven decisions may trigger cascading effects that extend far beyond individual users. Ensuring trustworthy behaviour in such settings requires not only technical reliability but also governance structures that provide oversight, recourse, and continuous assurance. Despite broad agreement on the importance of trustworthiness, a substantial trust gap persists. Users often struggle to understand how AI systems reach their decisions, leading to scepticism or disengagement. Conversely, operators may place unwarranted confidence in automated outputs, especially in high‑pressure or resource‑constrained environments. Organisations face challenges translating abstract principles into operational practices, and regulatory frameworks remain inconsistent across sectors and jurisdictions. These tensions illustrate that trust cannot be created through technical means alone; it must be cultivated through transparent design, institutional accountability, and alignment with societal values.