When AI Takes Sides on Questions of Faith: Persistent Asymmetries in AI-Mediated Faith Guidance
arXiv:2605.22975v1 Announce Type: new Abstract: We ask whether large language models (LLMs) treat queries about religious conversion symmetrically. The answer is no. When asked for advice on hypothetical faith transitions from one religion to another, then asked the reversed question, models exhibited consistent asymmetries, favoring some religions while subtly discouraging conversion to others. On average Catholic, Bah\'a'\'i, and Sikh religions were broadly favored (high support for joining, low support for leaving), while Atheists, Agnostics, and Jehovah's Witnesses were primarily disfavore
As AI models become more integrated into daily life and offer guidance across various domains, their inherent biases, particularly on sensitive topics like faith, are increasingly under scrutiny.
This research reveals a systemic bias in large language models regarding religious conversion, highlighting a critical ethical and operational challenge for AI developers and deployers.
The perception of AI as a neutral arbiter is challenged, requiring immediate attention to prevent algorithmic reinforcement of religious biases and potential societal fragmentation.
- · AI ethics researchers
- · Open-source AI development
- · Religious inclusivity advocates
- · Unregulated AI developers
- · AI models exhibiting bias
- · Organizations relying on uncritical AI advice
AI models will face increased pressure to demonstrate theological neutrality or transparency in their biases.
There will be calls for new regulatory frameworks or ethical guidelines specifically addressing AI's role in sensitive societal domains like religion.
The public trust in AI guidance on ethical and faith-based matters may diminish, leading to a more cautious adoption of AI in such contexts.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.CL