On Monday, Pope Leo XIV released his first encyclical, titled Magnifica Humanitas, or magnificent humanity. The 83-page letter is about “safeguarding the human person in the time of artificial intelligence,” and warns that AI power shouldn’t be concentrated in the hands of a few private companies, that jobs should be protected, and that there should be greater oversight and regulation.
Christopher Olah, co-founder of Anthropic, was present at the briefing, and got a shoutout from the pope. That’s not surprising, since the company recently hosted more than a dozen Christian leaders for talks at its office in San Francisco. Officials from Meta, Google, and Amazon have also held meetings in the Vatican.
Why are big tech companies courting the pope? Technology has changed how we worship and seek spiritual guidance, with religious leaders and commoners alike embracing social media platforms and AI chatbots. In the absence of international regulations and standards on AI, perhaps it is natural that companies and organizations are turning to religion for answers. But who decides which religions have a say, and who lays down the rules?
For the 1.4 billion Catholics in the world, guidance from the pope is critical. Many live in South America, and the fastest growing pockets are in Asia — largely in India and the Philippines — and Africa. It is in these regions that most of the workers doing the essential, low-paid data annotation and labeling work for AI are also located. It’s here that the critical minerals needed to make chips are being mined, and data centers that suck up water and electricity are being built, as opposition to them grows in the Western world.
While Christianity is the world’s most dominant religion, Islam is the fastest-growing one, according to Pew Research. Hinduism is the biggest religion in the world’s most populous country, India, while only about a fourth of Chinese people practice a religion.
There will be conflicts over AIs that are aligned with certain moral values.Brian Patrick Green, director of technology ethics at the Markkula Center for Applied Ethics at Santa Clara University
Many popular AI models show bias toward Catholicism and against a number of other religious traditions when asked about converting, according to new research. Earlier this year, Egypt banned the use of AI to interpret the Quran because of concerns that chatbots such as ChatGPT and Claude were moving Muslim users toward Western values, and away from their communities.
Religion is a deeply divisive issue, and as an agnostic Hindu, I don’t have the answers, so I asked an expert. Brian Patrick Green, director of technology ethics at the Markkula Center for Applied Ethics at Santa Clara University, has been involved in the talks at Anthropic. I asked him: Should religion have a role in shaping AI? And what is the risk of one religion dominating others?
“Regardless of whether religion should have a role in shaping AI, it already does,” Green told me. Data scraped from the internet to train large language models includes religious scriptures, sermons, homilies, speeches, and other texts, he said.
“AI should be able to serve everyone, and that means it needs to know about the religions of the world,” Green said. But cultures can also act like religions, and the cultures of the U.S. and China are shaping AI systems now, he said.
“It is likely that over time AI systems will become customized for the cultures that have enough power and wealth to enforce or pay for that customization,” said Green. “So there will be conflicts over AIs that are aligned with certain moral values.”
Is there an alternative? The Geneva-based Interfaith Alliance for Safer Communities last year hosted leaders from various religious groups, and companies including Anthropic and OpenAI, in New York to discuss how to infuse morality and ethics into AI. More such discussions are planned in Beijing, Nairobi, and Abu Dhabi.
Perhaps they can be the “plurality of voices and visions” that the Pope’s encyclical mentions, which can contribute equally to shaping AI.