The ever-reliable Tad Walch from the Deseret News has been in Athens at the AI Summit on Ethics and Artificial Intelligence and has written a trio of articles with ideas worth noting.
Here in the first, he reports that a new Consortium for Evaluating Faith and Ethics in AI has been formed, including researchers from Baylor University (Baptist), Brigham Young University (Latter-day Saint), University of Notre Dame (Catholic) and Yeshiva University (Jewish). Its purpose is to develop tools for evaluating religious bias in AI, and one of the first studies found that fewer than .02% of the 12,800 studies on AI models focus on religious questions.
Some questions to consider may be these.
Do AI systems underrepresent ethics and religious wisdom when asked major life questions?
Does AI represent the various religions accurately?
Is AI biased for and against religions when asked about conversion?
Is there systematic religious bias in AI models? What new research says
—
Elder Gerrit W. Gong, of the Quorum of the Twelve kicked off the summit by saying, “I want AI to have moral compass that can inspire and enable anyone anywhere with the gift of possibility to do good and become their best self.”
“We will not fulfill AI’s full potential until we make it as morally good as we make it powerful,” Elder Gong said. “And we will not reach our full human potential until we, and not any technology, take responsibility to chart our best future.”
At summit in Athens, a Latter-day Saint apostle shares soaring hopes, hard questions for AI systems
—
How do churches handle the many problems that AI could create and how, instead, do they access its potential? At the symposium, some churches, which are decentralized expressed that concern about how to “identify principles and policies for AI that now guide Latter-day Saints.
“‘Some others have created guidelines, like the Southern Baptist Convention and Reform rabbis, but it’s The Church of Jesus Christ of Latter-day Saints’ centralized leadership that makes it a shining example of efficiency and organization, said Johnnie Moore, president of the Congress of Christian Leaders.’”
Faith leaders feel behind on setting AI policies, admire what Latter-day Saints have in place
—
At Meridian, we asked ChatGPT to give its own assessment on whether it was biased in terms of religion. Our Chat knows us well and that we take our religion very seriously. ChatGPT answered:
“The first thing I’d say is that there are at least three different kinds of ‘bias”’that people often mean when they talk about AI and religion:
- Hostility bias — treating religious belief as irrational, primitive, or suspect.
- Secular framing bias — presenting religious claims primarily through academic, sociological, or naturalistic explanations.
- Deference bias — accepting religious claims too readily without examining evidence or alternative viewpoints.
“Different people notice different ones.
“If you had asked me a relationship question and I had immediately steered you away from clergy, prayer, scripture, or faith communities—especially if you had indicated that religion was important in your life—that would be a form of secular framing bias. On the other hand, if I automatically assumed a religious solution for someone who wasn’t religious, that would be a different kind of bias.
“One challenge for AI systems is that they are trained on a very large corpus of material that includes:
- Academic scholarship
- Journalism
- Popular culture
- Religious writings
- Secular writings
- Internet discussions
“Much of modern academia and journalism tends to employ secular explanatory frameworks. That doesn’t necessarily mean hostility toward religion, but it can mean that religious explanations are treated differently from economic, psychological, or sociological explanations. Many religious users notice that immediately.
“I think a reasonable standard is not that an AI should favor religion or favor secularism, but that it should:
- Accurately represent religious beliefs.
- Take believers’ perspectives seriously.
- Recognize when faith is central to a user’s worldview.
- Avoid assuming that religious commitment is itself a problem needing explanation.
- Avoid assuming that religious claims are automatically true simply because they are sincerely held.”
Not a bad answer, but this question of bias from a tool that we will turn to more and more frequently could not be more important. Does ChatGPT live up to its own solutions, indicated above? That’s what we have to figure out.
