Da’wah in the Algorithmic Era: Investigating Bias and Validity of Islamic Artificial Intelligence Applications

Abstract

The emergence of Islamic AI applications represents a pivotal transformation in the mediation of daʿwah. Yet, it raises pressing concerns about bias, epistemic validity, and the absence of recognized religious authority. This study examines ChatMu GPT and MuslimAI.ai, two prominent platforms with distinct theological orientations, to evaluate how they address religious questions framed within Indonesia's four indicators of religious moderation across the domains of message of da'wah, such asʿaqīdah (creed), ʿibādah (worship), and muʿāmalah (social ethics). Using qualitative content analysis and drawing on Algorithmic Bias Theory, six prompts were designed to test inclusivity, doctrinal balance, and ethical framing. Findings reveal that ChatMu GPT consistently grounds responses in Muhammadiyah doctrinal sources, producing structured but institutionally aligned guidance, while MuslimAI.ai prioritizes emotional inclusivity and accessibility at the expense of jurisprudential depth. Both exhibit limited transparency of sources and potential algorithmic bias. These results suggest that while Islamic AI can enhance accessibility to religious knowledge, its unsupervised use risks narrowing interpretive diversity. The study recommends hybrid AI models supervised by qualified ʿulamāʾ, trained on pluralistic datasets, and aligned with Islamic ethical principles of ṣidq (truthfulness), amānah (trustworthiness), and maṣlaḥah (public good) to preserve the integrity of digital daʿwah.