“The Prophecy of Machine?”: Exploring Ethical Issues embedded in AI-Driven Social Simulacra Research

Abstract (English)

This study examines the ethical consequences of AI-driven social simulacra in anthropological research, exemplified by Stanford’s AI Town, which transforms cultural phenomena into adjustable parameters for controlled experimentation. Although generative AI advances agent-based modeling and ethnographic methodologies(Epstein, 2006; Piao et al., 2025), ethical governance remains underdeveloped. Based on media evocation(Van der Goot & Etzrodt, 2023) and materializing morality(Verbeek, 2006), this research employs an ethical impact assessment framework to critically examine the emergent ethical challenges arising from anthropological research under AI-driven simulacrum. Ethical problems emerging from AI-driven simulacra in anthropological research include: 1)AI turns researchers’ subjective norms into opaque algorithms, freezing social and cultural complexity into static variables that risk naturalizing power hierarchies as objective truths. 2)A large number of manipulation of simulated agents blur ethical boundaries around informed consent and moral consideration of AI agents’ emergent behaviors. 3)Traditional ethical principles like beneficence and justice falter with infinitely replicable digital subjects and compressed simulation timelines, while feedback loops from simulation to policy create new ethical entanglements. Additionally, AI simulacra undermine cultural agency through reified interaction models, potentially exacerbating epistemic inequality by commercialized platforms. Current ethics frameworks, rooted in physical-world interactions, fail to address the ontological challenges of AI-generated anthropological realities. The study argues for situated ethics that embed pluralistic values in simulation design while preserving space for unpredictable social dynamics to counter algorithmic determinism. This intervention holds urgent relevance for anthropological studies as AI-driven methodological innovations risk perpetuating digital epistemological violence without interdisciplinary governance recognizing simulations as ethical artifacts.