Plural Interfaces, Singular Commitments: Religious Diversity, AI Alignment, and the Non-Collapse of Output, Belief, and Moral Responsibility

Abstract

Conceptual article in philosophy of technology / AI ethics / Islamic ethics. Public-facing AI systems mediate religious questions across societies with incompatible metaphysical commitments; observers often collapse five distinct objects into one: model output, product policy, institutional normative commitment, an actor's interior belief, and a person-specific religious classification. This article distinguishes interface pluralism from metaphysical relativism, develops an Islamic normative architecture (tawḥīd, taʻāruf, faqr, amāna, taklīf), and proposes the Religious Attribution Non-Collapse (RANC) principle together with a role-sensitive Role–Amāna responsibility audit. v1.0, journal-submission draft (not peer reviewed); target journal Philosophy & Technology. The manuscript's one formally stated object (RANC, sec.8.1) is registered in the Toledo equation library (morrocwi/toledo) as code weld/H.51.v1.