Agency Attributions and AI Agency

Document Type : Research Paper

Author

Imam Khomeini Institute

Abstract

The contemporary debate on AI agency is constrained by an essentialist focus on whether artificial systems possess genuine autonomy or traditional belief-desire agency. This paper argues that such an approach constitutes a category error, obscuring realities of AI deployment. Instead, it proposes an attributionist framework that understands agency as emerging from socio-technical relationships rather than residing exclusively in algorithms.

Drawing on the theological concept of *fā‘il biltajalī*, or agency through mediated manifestation, the paper conceptualizes AI systems as conduits of human intentions and institutional structures. This perspective offers an alternative to prevailing technical approaches. By synthesizing Sven Nyholm’s analysis of responsibility gaps with Herman Veluwenkamp’s distinctions, the paper develops a precise accountability taxonomy. Specifically, it distinguishes Epistemic Responsibility Gaps (ERG), arising from system complexity and limited explainability, from Control Misalignments (CM), resulting from institutional delegation failures.

Ultimately, this paper reframes the Alignment Problem as a socio-technical challenge of normative integration rather than a technical task of matching objective functions to predefined goals. An aligned AI system is not necessarily one exhibiting independence, but one operating within legible, revisable, and accountable boundaries of human institutional intent. The attributionist perspective thus shifts attention from machine autonomy to distributed responsibilities underpinning AI deployment.

Keywords

Main Subjects

Send comment about this article
CAPTCHA Image
Journal of Philosophical Theological Research

Articles in Press, Accepted Manuscript
Available Online from 03 October 2026

Files

Share

How to cite

Statistics