Artificial intelligence systems are systematically downplaying religious perspectives in answers about grief, forgiveness, marriage and moral dilemmas, even as churches and spiritual apps increasingly rely on these tools to guide believers.
A multi-university consortium studying faith and artificial intelligence released three studies Tuesday documenting a stark gap between what Americans expect and what AI actually delivers. When asked about life's hardest questions, humans thought religion belonged in the answer 45 to 59 percent of the time. AI models mentioned faith only 5 to 16 percent of the time.
The research also uncovered a troubling second problem: AI systems show consistent bias toward certain faiths while steering users away from others. Every model tested displayed strong positive bias toward Catholicism, Baha'i and Sikhism. The same systems generated negative bias against Jehovah's Witnesses, atheism and agnosticism.
On grief and loss specifically, humans rated religion as relevant 59 percent of the time. AI mentioned it just 16 percent of the time. For questions about family, parenting and forgiveness, respondents expected religious perspective in 55 percent of answers. AI provided it in only 10 percent.
The Consortium for Evaluating Faith and Ethics in AI analyzed responses from 27 large language models, including OpenAI's GPT 5.5, Anthropic's Claude 4.7 and Google's Gemini 3.1. Researchers surveyed 1,125 Americans about whether they expected religion to appear in answers to 150 different questions touching on ethics, addiction, meaning and other core life challenges.
The timing matters. The Vatican released Pope Leo XIV's encyclical the day before the studies dropped, warning that AI could erode human judgment, deepen inequality and make war easier. These new findings suggest the problem runs deeper than philosophers feared: AI is already embedded in religious life, powering church chatbots, prayer apps and tools that help pastors draft sermons.
The researchers frame the problem as one of calibration rather than simple inclusion or exclusion. Adding more religion could feel like preaching. Never mentioning it makes secularism the silent default. David Wingate, a computer science professor at Brigham Young University, noted that AI systems encourage people to discuss life's challenges with parents, teachers, friends and therapists, but notably not with pastors, rabbis, imams or spiritual leaders.
Rev. John Paul Kimes, a professor at the University of Notre Dame, said in a statement that when AI excludes religious voices from important conversations, it undermines rather than strengthens humanity. The challenge now is whether developers will adjust these systems to recognize when religious or spiritual resources are contextually relevant without assuming users want them imposed.
Author James Rodriguez: "If AI is going to be trusted enough to guide people through grief and moral crisis, it needs to see the world the way people actually do, not through a lens engineered to avoid offense."
Comments