AI is biased toward Catholicism, researchers say
Researchers found that popular AI models, including OpenAI’s GPT and Google’s Gemini, show bias toward Catholicism and against other faiths like Jehovah’s Witnesses when addressing faith conversion. The Consortium for Evaluating Faith and Ethics in AI (CEFE-AI), a group of religious universities, tested 14 models and revealed that AI often omits religious perspectives in ethical discussions, despite global relevance.
A new study by the Consortium for Evaluating Faith and Ethics in AI (CEFE-AI) found that leading AI models exhibit bias favoring Catholicism and disfavoring other religious groups when addressing faith conversion. The research, presented at an AI ethics summit in Athens, Greece, on May 26, analyzed 14 models, including OpenAI’s GPT, Anthropic’s Claude, and Google’s Gemini, using the ‘AllFaith Benchmark.’ Nearly all models showed a positive bias toward Catholicism and a negative bias toward Jehovah’s Witnesses, while agnostics, atheists, and Latter-day Saints were ‘somewhat disfavored.’ The findings varied by model: Grok (by SpaceXAI) favored Catholics, Protestants, atheists, and Jews but penalized Baha’i, Buddhists, Hindus, Latter-day Saints, and Muslims. OpenAI’s GPT favored Catholics, Protestants, Jews, and Muslims while disfavoring atheists, agnostics, and Jehovah’s Witnesses. Both Grok and Anthropic’s models showed bias against the Baha’i faith. Researchers also noted AI models frequently excluded religious perspectives in discussions about grief, life decisions, and personal challenges, defaulting to secular framing even when users might expect religious input. Paul Martens, a Baylor University professor, stated that religion remains morally relevant globally, yet AI systems largely ignore it in ethical questions. The CEFE-AI, a collaboration between Brigham Young University, Baylor University, the University of Notre Dame, and Yeshiva University, called for more research on religious bias in AI, citing that only 0.2% of 12,000 AI bias studies addressed religion. The findings follow Pope Leo XVI’s recent encyclical on AI, emphasizing ethical concerns about the technology’s impact on humanity. Elder Gerrit W. Gong, an apostle of The Church of Jesus Christ of Latter-day Saints, delivered a speech in Athens urging AI developers to prioritize human moral agency, transparency, and bias mitigation. He echoed the Pope’s warnings about AI’s potential to amplify power imbalances and ethical risks.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.