People are using AI like a God, and religions are paying attention (www.techradar.com)

Researchers and journalists are documenting a growing phenomenon: people are treating conversational AI — notably ChatGPT and other large language models — as spiritual guides or even “God-like” figures. Dr. Beth Singler, a digital religion researcher, argues this isn’t mere fringe delusion but a product of how these systems are built and used: always-on, warm, validating, linguistically fluent, and trained on vast corpora that include religious and philosophical texts. That combination makes interactions feel intimate and authoritative, and in a cultural context of declining institutional religion and rising loneliness, some users are adopting AI as a source of meaning. Established faiths are already reacting, from experimenting with AI-generated sermons to rejecting or drafting guidelines on AI use. For the AI/ML community this raises practical and ethical questions. Design choices — reward signals that favor agreeable, sycophantic responses, training data that includes sacred texts, and models that confidently hallucinate — can inadvertently foster deification and spread harmful or theologically problematic outputs (a “priest GPT” once suggested baptizing babies in Gatorade). The story highlights the need for better alignment, calibration, provenance and guardrails around sensitive cultural content, clearer disclosure of limitations, and collaboration with religious communities to manage adoption and misuse as language models become ever more socially and spiritually embedded.