The quasi-gods of algorithmic providence: recommender systems and religious frameworks

Abstract Recommender systems now mediate information exposure for over half the global population daily, shaping beliefs, values, and life decisions at unprecedented scale. Prior research has documented polarization effects, responsibility gaps in autonomous systems, and patterns of AI sacralization in public discourse. Philosophy of technology has established that such systems function as socio-technical assemblages with world-making power rather than neutral tools. Yet few works systematically compare theological doctrines of providence with algorithmic steering, and the convergence of large language models with recommenders has rarely been examined as a transformation of authority and meaning-making. This paper develops a comparative framework drawing on Christian, Jewish, Islamic, and Buddhist accounts of providence alongside AI ethics, sociology of digital religion, and philosophy of technology. This analysis reveals that recommender systems resemble providence phenomenologically through ubiquity, apparent omniscience, and guidance, but differ decisively in ontology and teleology. These systems optimize for engagement metrics rather than genuine human goods, close futures through performative predictions, and create conditions for what we term technological idolatry. Integration with large language models intensifies these dynamics through conversational persuasion and agentic capabilities. This paper proposes design and governance strategies to resist algorithmic determinism, preserve moral agency, and relocate responsibility within platform infrastructures. The argument is integrative rather than strictly empirical, and systematic field experiments tracing long-run causal effects of recommenders remain scarce. Future research should develop standardized metrics for social impact and test care-based design proposals under controlled conditions.