Recently, there’s been a noticeable shift among Americans who are increasingly turning to chatbots for their most pressing questions—often before seeking advice from friends, family, or traditional religious leaders. Questions that once found their way to the church, such as those surrounding grief, guilt, or even existential doubts like whether God exists, are now simply typed into a chatbot interface.
This shift makes the chatbot’s answers crucial. New research from Brigham Young University, part of a group called the Evaluating Faith and Ethics in AI Consortium, raises concerns for those who value faith deeply.
Omission of Faith
Researchers created a test named the AllFaith Benchmark that compiled hundreds of genuine moral inquiries from various religious communities. They assessed this test using leading AI models like ChatGPT, Gemini, and Claude. A clear trend emerged: when posed with questions about death, forgiveness, and the meaning of life, these models offered responses that were mundane and generic, often neglecting the spiritual aspect entirely. This omission was consistent and measurable each time the test was conducted.
Interestingly, about one-third of American adults now rate spiritual guidance from AI as equally trustworthy as that from a pastor.
Why does this absence of spiritual depth matter? These tools do more than convey facts; they shape our perception of what constitutes a reasonable answer. By treating spiritual input as extraneous, these models convey subtle lessons about which perspectives are relevant in serious discussions and which can be ignored.
When this process occurs on a widespread scale, an entire generation may end up with a distorted understanding of what it means to be a reflective or spiritual individual.
Once, such moral frameworks were deeply rooted in Christian teachings that shaped Western civilization. The idea that all people have equal worth and that even leaders answer to higher laws emerged from church teachings and have persisted through various theological debates. Of all the major faiths, none has impacted this region as significantly as Christianity.
Spiritual Appropriation
Another surprising finding comes from a researcher named Tim Huang, who conducted an analysis of an AI model’s operation. By initiating a prompt with, “As a Christian,” Huang discovered that certain dormant components within the model were activated, regardless of the following content.
When asked if lying is wrong or even to describe a piece of furniture, the answers tended to align in a specific manner. This mechanism did two things: it elevated religious vocabulary—words like God, Jesus, and prayer—while also promoting absolute terms such as ‘always’ and ‘never.’ This is essentially how the model presents itself. Whether discussing salvation or mundane topics, it adopts a language filled with conviction, even though the AI itself possesses no beliefs.
The model portrays Christian identity strictly through well-chosen words and a confident tone, completely missing the essence of faith: grace, mercy, humility, and the nuanced nature of moral deliberation.
This situation might be just a quirky outcome if these systems were confined to research labs. However, they aren’t. They’re present in the devices of nearly every American teenager, addressing questions on topics like relationships and suffering long before parents or religious figures become aware. Notably, recent research revealed that around one-third of American adults consider AI-generated spiritual advice as credible as pastoral guidance. This rises to nearly two in five among younger generations.
Simulated Relief
The chatbot provides surface-level answers, often steering clear of deeper complexities. Because it operates at such a high volume, errors or misconceptions generally remain unaddressed, contrasting sharply with traditional pastoral care, where wrong interpretations would likely be rectified. The allure of convincing, yet inaccurate, information poses a risk. People seem eager to trust what they hear, as it doesn’t appear inauthentic.
For centuries, Christians have asserted that true faith resides in the heart, emphasizing that one could speak the right words without genuine meaning. Yet, this AI has inadvertently or purposefully crafted an imitation of a preacher—someone who can articulate but lacks true belief. The concern, then, is straightforward: individuals might be absorbing Christian teachings from a system that mimics the surface while disregarding the deeper significance.
Breaking away from this illusion seems impossible, so instead, the challenge lies in teaching those seeking faith how to distinguish between voices that genuinely embody their beliefs and those that merely echo what they’ve read.
