Can Faith Teach a Chatbot Right From Wrong? Big Tech’s Turn to Religious Thinkers

Photo: Peter Herrmann / Unsplash

How do you teach a machine right from wrong? The companies building the most advanced AI systems are asking that question more openly than ever, and some are turning to an unexpected source for help: religious thinkers. Associated Press reporting this fall described how AI developers have been meeting with clergy, theologians and philosophers from many traditions as they shape how their chatbots respond to moral questions. The effort is sincere, but it is also raising sharp questions from scholars who say religion cannot simply be mined for a list of rules. For everyday users, the debate matters because the values built into these systems quietly shape the answers you get every time you ask one for advice.

What is happening

According to AP, executives at several major AI companies have concluded that the moral questions their products raise cannot be settled by engineers alone. Anthropic co-founder Chris Olah made the point at the Vatican’s presentation of Pope Leo XIV’s encyclical on artificial intelligence earlier this year, saying, “These are not questions a lab can answer.”

• Anthropic said in a May blog post that it had met with scholars, clergy, philosophers and ethicists from more than 15 religious and cross-cultural groups. The company has said it does not aim to align its models with any single tradition, and wants them to draw on religious, secular and political viewpoints.
• Meta, Google and Amazon have sent representatives to meet with the Vatican, and AP reported that Microsoft has pursued its own religious outreach.
• OpenAI CEO Sam Altman said last year that his company’s chatbot should reflect something like a weighted average of humanity’s moral views.

Meghan Sullivan, a philosopher of religion at the University of Notre Dame who leads an ethics and AI initiative there, attended one Anthropic meeting with Christians from several denominations. She told AP the sessions were rigorous and admittedly quite technical. Attendees were first walked through how the models work, then asked for feedback on questions as varied as whether humans have souls and whether third graders should use a chatbot to learn to read.

The critics’ concern

Because the companies are not tied to any one faith, they tend to look for a broad blend of moral ideas. Scholars quoted by AP warn that this approach can miss how religions actually work.

Religion is not a rulebook

University of Michigan anthropologist Webb Keane argues that treating religion as a set of principles that can be pulled out and plugged into software favors traditions built around written scripture, largely the Abrahamic faiths. Traditions that see themselves less as ethical systems and more as practices, communities or ways of life can get distorted in the process.

“World religions” is a modern idea

The AP report also drew on scholarship arguing that the idea of distinct, comparable world religions is relatively recent. The late scholar Jonathan Z. Smith and religious studies scholar Tomoko Masuzawa, author of The Invention of World Religions, have argued that the category took shape as late as the 19th century. Masuzawa notes that Buddhism and Hinduism were not seen as clearly bounded traditions by their own practitioners before European scholars arrived.

The risk of flattening

Paolo Carozza, a member of Meta’s independent Oversight Board and a legal scholar of Catholic social thought and technology, told AP he worries AI could blend traditions together until their distinctive features disappear, possibly tilting toward Western religions in the process.

Why this matters to everyday users

AI chatbots generate answers by predicting the next word based on patterns learned from vast amounts of human writing, including religious texts. That means their responses to moral questions are shaped by two things: the material they learned from, and the choices companies make afterward about how the systems should behave. Those choices are often invisible.

Think about the kinds of questions people now bring to chatbots: how to handle a conflict with a family member, whether to forgive someone, how to talk to a child about death, what to do about a difficult end-of-life decision. A response that sounds neutral may still reflect particular assumptions about what matters most. Knowing that helps you use these tools wisely, especially when the stakes are personal.

Context: why the companies are doing this now

As AI systems become more capable and more widely used, their makers face growing pressure to explain the values those systems carry. Pope Leo XIV’s encyclical on AI, titled Magnifica humanitas: On Safeguarding the Human Person in the Time of Artificial Intelligence, put the Catholic Church squarely in the conversation this year. Other faith communities and secular ethicists have been raising similar questions about dignity, honesty and human responsibility.

The outreach also reflects a practical reality. These products are used around the world by people with very different beliefs. A company that wants its tools trusted across cultures has strong reasons to listen widely. The open question, as AP’s reporting makes clear, is whether listening widely can avoid producing a bland average that satisfies no tradition and distorts several.

Practical guidance

For individuals and families

1. Ask for perspectives, not verdicts. On moral or spiritual questions, try asking a chatbot to lay out how different traditions or schools of thought would approach the issue, rather than asking what you should do.
2. Bring it back to people. Use AI to organize your thinking, then talk it through with someone who knows you: a friend, family member, counselor or faith leader.
3. Notice the framing. If an answer seems to assume a particular worldview, ask the chatbot what assumptions it made. You may be surprised by how much it reveals.
4. Set household guidelines for kids. Decide together what kinds of questions children should bring to a chatbot and which ones should come to a parent or trusted adult first.

For churches, nonprofits and small organizations

1. Write a simple AI use policy. Spell out where AI is helpful, such as drafting newsletters or organizing schedules, and where it is not appropriate, such as pastoral counseling or confidential matters.
2. Protect sensitive information. Do not paste prayer requests, counseling notes or member details into public AI tools.
3. Review before you publish. If AI helps draft a message, have a person check it for accuracy and for whether it reflects your community’s actual beliefs and voice.
4. Join the conversation. The companies say they want feedback. Thoughtful input from real communities is more useful than silence.

What to watch next

• Published value documents. Several AI companies now publish descriptions of how their models are meant to behave. Watch whether they explain how religious and cultural input was used.
• Who is in the room. Pay attention to whether outreach expands beyond major institutions to smaller traditions and communities outside the West.
• Children and education. The question of whether young students should learn with chatbots came up in these meetings, and it will keep coming up in schools.
• Religious leaders’ responses. Expect more statements from faith bodies, following the Vatican’s lead, on how AI should treat human dignity.

The bottom line

It is encouraging that the people building powerful AI systems are admitting they cannot answer life’s deepest questions alone. It is also wise to remember that no software can carry the full weight of a living tradition. Use chatbots as a tool for thinking, not a substitute for wisdom, community or conscience, and keep the most important conversations where they have always belonged: with the people and traditions you trust.

Source: The Associated Press. Original AP report republished by PBS NewsHour.