Aswaja An-Nahdliyah in the Age of Generative AI: An Ethical-Epistemological Framework

Abstract

Generative artificial intelligence can accelerate knowledge work while increasing exposure to fabricated sources, misplaced confidence, representational bias, and uncertain scholarly authority. This conceptual study reconstructs Aswaja An-Nahdliyah as a situated ethical-epistemological framework for assessing AI-assisted knowledge production. An integrative, non-systematic review connects research on language-model risks, responsible AI, Islamic ethics, and Nahdlatul Ulama’s normative orientations. Comparative interpretation and negative-case reasoning yield four interdependent checkpoints: tawassuth assesses the purpose and proportionality of AI use; tawazun establishes meaningful human interpretive control; i‘tidal evaluates representational justice and attributable responsibility; and tabayyun requires evidence tracing, claim assessment, contextual interpretation, and publication-risk review. Together they constitute the Aswaja Ethical-Epistemological Framework for Generative AI (AEEGA). Its proposed novelty does not consist in discovering Islamic AI ethics or inventing a new verification obligation. Rather, it couples culturally grounded normative reasoning with an inspectable sequence that distinguishes generated content, justified claims, and authorized scholarly judgment. Compared with existing Islamic virtue-based, trusteeship, Aswaja cyber-ethics, and general AI-risk approaches, AEEGA specifically addresses the transition from probabilistically generated religious or academic statements to accountable knowledge claims. A worked hypothetical case illustrates possible institutional application. The framework remains a theoretical proposition, not an empirically validated instrument; its discriminant validity, usability, and effects require testing with scholars, educators, and students. This contribution supports pluralist AI governance without replacing technical audits, qualified interpretation, or legal protection.