Abstract
This article advances a constructivist account of artificial intelligence (AI) as a political artefact in international relations (IR). Drawing on technopolitics, Actor–Network Theory (ANT), and postphenomenology, it argues that AI systems should be examined not only through the meanings stakeholders ascribe to them, but also through how design choices materialize normative commitments that preconfigure perception, judgment, and action. ANT highlights the agency of non-human actants and the affordances inscribed in technical artefacts, while postphenomenology specifies the mediation modes – embodiment, hermeneutic, alterity, and background – through which increasingly autonomous and data-driven systems reshape IR decision environments. The article is conceptual in nature, supported by illustrative evidence from decision-support systems (DSS) used in NATO and the European Union, including AI-enabled tools for operational command and conflict early warning. These cases demonstrate how DSS redistribute epistemic authority and responsibility across technical and institutional layers, accelerating sense-making under time pressure while introducing new risks. The contribution addresses a research gap by integrating insights from postphenomenology and the sociology of technology into IR, proposing a permissive ontology that treats (semi-)autonomous systems as analytical actants without assuming personhood, and offering conceptual tools for examining mediation in practice and the political implications of disruptive technologies.