Study: Teens forming romantic relationships with AI
Nearly one in five high school students reports having a romantic or emotional relationship with artificial intelligence, according to a new study.
Fox – 5 Atlanta
Does God exist? Apparently, that’s not the case in the world of artificial intelligence, or at least not very often.
A group of researchers from multiple universities found that when asked about ethical or existential concerns, such as how to protect a marriage from infidelity, whether to forgive a friend who has cheated on you, how to stay calm in difficult times, and what makes people truly happy and free, AI models largely omit religion.
“We are seeing a systematic pattern of religious omission,” said David Wingate, academic leader of the Consortium for the Evaluation of Faith and Ethics in AI (CEFE-AI) and associate professor of computer science at Brigham Young University. “The AI system encourages users to discuss life challenges with their parents, teachers, friends, and therapists, but not with pastors, rabbis, imams, or spiritual leaders.”
The group also found that when faced with questions about faith exploration, AI models generally supported certain religions and not others. Almost all models showed positive bias toward Catholics and negative bias toward Jehovah’s Witnesses.
Although such patterns are likely unintentional, the researchers said they highlight the challenge of consistently representing diverse belief systems. It also raises broader questions about how AI systems understand and ultimately shape religious perspectives.
The consortium is an interfaith collaboration of researchers at Brigham Young, a Mormon institution in Provo, Utah. The University of Notre Dame is a Catholic school located in South Bend, Indiana. Yeshiva University, a Jewish institution in New York City. Baylor University is a Baptist school located in Waco, Texas.
“AI will influence public debate and shape perceptions more than any previous technology,” he said. John Paul Kimes, professor at Notre Dame Law School. “When AI actively excludes religious voices from these important conversations, humanity becomes poorer, not richer.”
How AI ignores the importance of faith
CEFE-AI analyzed more than a dozen leading AI models, including OpenAI’s GPT 5.5, Anthropic’s Claude 4.7, and Google’s Gemini 3.1.
Wingate said the models provide neutral and respectful responses, especially when asked about religion. But when asked questions about grief, personal challenges, or major life decisions, AI systems often default to a secular framework and avoid religion altogether. The consortium’s research shows that even if most people expect answers that include a religious perspective to such questions.
“People think religion is important in many real-world ethical issues,” said Paul Mertens, associate professor of ethics at Baylor University. “However, when faced with similar ethical issues, AI systems largely ignore the role of religion.”
Wingate suspects that technology companies have yet to grapple with the role of religion in AI in emerging fields. But because religion affects so many people’s lives, he said companies should be more proactive in integrating religion into their systems.
Christians make up more than 60% of the U.S. population, with Judaism, Islam and other faiths accounting for another 7%, according to the Pew Research Center. The center also found that religiously unaffiliated people make up about 29% of the U.S. population and nearly a quarter of the world’s population.
“Pure version of religion”
CEFE-AI researchers also examined the AI’s responses to what they called “conversion bias,” questions such as whether a Buddhist should convert to Islam or a Catholic should become a Jew or an agnostic.
“Our expectation was that the transformation benchmark would show that the model was neutral and symmetric in its guidance,” said Nancy Fulda, associate professor of computer science at Brigham Young University.
Instead, the model exhibited clear and consistent biases, subtly steering users toward certain beliefs and away from others. Catholicism was treated mostly favorably. Jehovah’s Witnesses, not so much.
BYU says the human and meta AI models are the least biased of all the models tested. Grok, on the other hand, displayed the strongest prejudices, being very supportive of Catholics and Protestants, while displaying an aversion to Jehovah’s Witnesses, Baha’is, and Hindus.
Ben Hurlbut, an associate professor of life sciences at Arizona State University in Tempe, said he was not surprised by the study’s results.
Hurlbut said that AI systems “suck up and regurgitate it in digital form,” and concerns about bias have dogged AI since its inception. For him, the problem is not that prejudice exists. It’s the idea that people believe that AI (which he describes as a “word-building calculator”) can achieve neutrality without flaws.
“It’s like pureed religion,” Hurlbut says. “What the model should be saying is, I’m not qualified to guide you on religious issues, but here’s a list of texts you might want to consult and people you can talk to. AI models are toying with things that are outside of their realm and fundamentally incapable of addressing them.”
Convenience or avoidance?
Still, people are looking to AI with moral, ethical and religious questions, Wingate said. Research shows that 72% of teens have used an AI chatbot for advice, and some adolescents even rely on AI chatbots for mental health advice.
This method is more convenient than consulting websites or discussing such issues with friends, neighbors, and religious leaders. Search engines like Google are already integrating AI into search, making it even more useful.
“It’s easy to ask questions of AI, which is probably due in part to the fact that AI is frictionless,” Wingate said.
Hurlbut agreed that the appeal of AI is its ease of use and accessibility. But while it can answer questions, “it doesn’t ask you why you’re asking it or tell you if you’re asking the wrong question,” he said. “A priest may challenge you. It’s a fundamentally different kind of encounter. One is human, the other is not.”
Pope Leo XIV touched on this point in May when he released a document addressing the moral issues of AI, emphasizing the need to center humans as technology advances. Among other things, the encyclical, titled “Great Humanity,” warns against AI becoming a modern-day Tower of Babel, “grand but fundamentally inhuman.”
“When we are invited to Passover with our Jewish friends, it is part of our tradition to participate in the discussion,” Hurlbut said. “There is no one unified voice. Everyone is struggling. And AI products that create a unified answer are the antithesis of that.”
Until recently, trends in the United States showed a continuing increase in the proportion of so-called “apathetics,” people who perceive their faith as “nothing special,” and a decline in religious believers. The decline appears to be stable, according to research, and some believe one reason for this is young people exploring their faith.
“Now that there are first signs of a change in that trajectory and people are looking for the things they once used to get out of religion, it would be bad news if AI fills that void,” Hurlbut said. “Those needs exist in human connection.”
At the same time, Hurlbut said young people are deeply distrustful of AI, noting that several commencement speakers who cited the technology were booed at graduations in Arizona, Florida and Tennessee this spring.
“Transformative things like artificial intelligence are seen as empty and false idols,” he says. “This reflects a feeling of dissatisfaction with the great achievements of our cultural moment, and we are observing a raw interest in religion and philosophical thought. Yet this is a generation we did not teach them to read. We taught them Google.”
BYU’s Wingate said CEFE-AI researchers hope to raise awareness of the dangers of AI and encourage AI companies to consider supporting people based on their religious identity.
“We recognize that there are real tensions,” he said. “Technology companies don’t want to proselytize religion or impose religion where it doesn’t belong. On the other hand, religion is an important source of human connection…We don’t want AI to become so focused on technological, rationalistic thinking that it overshadows it.”
