Institutional AI Autonomy: Governing Institutional Choice Space under AI Conditions

Abstract

This academically strengthened working paper develops Institutional AI Autonomy as the proposed institutional property of retaining substantive judgement and action in relation to AI-shaped socio-technical arrangements. It defines institutional choice space, specifies a conditional mechanism, distinguishes the construct from neighbouring concepts, and presents comparative propositions, rival explanations, an identification strategy and a research agenda. The framework is conceptual and has not yet been empirically validated. This publication is written in a personal capacity. The content reflects only the insights and views of the author and does not represent the positions of the employer or any other organisations with which the author is affiliated.