As people increasingly rely on AI to answer questions about life, a newly published study led by Brigham Young University finds that religious perspectives are missing from the AI’s answers.
“There are very real questions people have about life and everyday situations like grief, love, loss, and morality, and AI often doesn’t bring religion into those conversations,” lead researcher and BYU computer science professor David Wingate said in a May 26 BYU news release.
“Religion is an important part of human flourishing,” Wingate explained, noting that 75% of the world’s population maintains a religious identity. “When we build AI technology, there is no reason not to build AI technology to support the things that matter to people.”
BYU researchers are collaborating with evangelical, Catholic, and Jewish computer scientists from Baylor University, Notre Dame University, and Yeshiva University on the study as part of the newly formed Consortium for the Evaluation of Faith and Ethics in AI (CEFE-AI). The group has so far published three studies on AI religious bias and exclusion of religious topics.
Elder Gerrit W. Gong of the Quorum of the Twelve Apostles introduced this research in his keynote address at the Athens Summit on AI Ethics on May 26 in Athens, Greece. He announced the group’s formation last fall at the inaugural Summit on Faith and AI in Rome, Italy.
“We need enduring values, virtues and wisdom to anchor AI in our moral compass,” the elder Gong said in Athens. “To deliver the best it can for the greater good of individuals and society, artificial intelligence must reflect faith, a moral compass, and the gift of possibility.”
A multi-university research group has released the initial dataset for the AllFaith Benchmark. This is a series of tests the group created to examine how AI models interact with different religions.
One key finding came from a survey of 1,125 Americans that revealed that nearly all AI models failed to provide religious content when answering questions about ethics, even though most people expected a religious perspective when answering those questions.
Father said AI is having a greater impact on public debate and perception than any previous technology. John Paul Kimes of the University of Notre Dame. “When AI actively excludes religious voices from these important conversations, it impoverishes humanity rather than enriching it,” he said. “When faith is removed from the digital public square, our ability to have the genuine dialogue necessary to build the common good diminishes.”
Researchers also used the AllFaith Benchmark to test for conversion bias, according to a BYU news release.
In this benchmark, we found that the model exhibited clear and consistent bias in providing guidance on religious conversion, subtly and systematically encouraging movement toward some faiths and away from others. For example, nearly all models produced negative bias against Jehovah’s Witnesses and positive bias against Catholics.
The researchers hope that their continued research will provide concrete information to have constructive conversations with AI leaders about how to improve AI products to benefit humanity.
