Trust, trustworthiness and AI governance.

Area of Science:

• * Artificial Intelligence (AI) Alignment
• * Public Governance and Algorithmic Decision-Making (ADM)

Background:

• * Growing integration of AI and ADM in core state functions presents alignment challenges.
• * Trust and trustworthiness are critical, yet complex, considerations in AI governance.
• * Existing scholarship is fragmented across sociology, political science, and computer science.

Purpose of the Study:

• * To provide an interdisciplinary overview of trust and trustworthiness in AI.
• * To argue for a comprehensive approach to understanding AI alignment in governance.
• * To propose a roadmap for addressing challenges in trustworthy AI-Governance.

Main Methods:

• * Interdisciplinary literature review synthesizing scholarship on trust.
• * Analysis of trust and trustworthiness properties across different academic fields.
• * Conceptual framework development for socio-technical AI contexts.

Main Results:

• * A coherent understanding of AI alignment necessitates integrating diverse trust properties.
• * Trustworthiness in AI-Governance requires simultaneous consideration of machines, humans, and institutions.
• * Complex watchful trust dynamics emerge in socio-technical AI applications.

Conclusions:

• * An interdisciplinary approach is essential for effective AI alignment in public authorities.
• * Addressing trust and trustworthiness is key to developing reliable AI-Governance systems.
• * A strategic roadmap is proposed to navigate the challenges of trustworthy AI integration.