Religious groups must keep careful eye on artificial intelligence

Programmers, computer scientists and software, mechanical, data and prompt engineers — these are some of the professions behind the creation of artificial intelligence. Should theologians and faith leaders also be involved?

Meghan Sullivan, a Roman Catholic who teaches philosophy at the University of Notre Dame, says yes. That’s why she was glad to attend a meeting in March at the invitation of Anthropic, the creator of Claude AI, about the role religion can play in the creation of this life-changing technology.

Sullivan, who also directs the university’s Institute for Ethics and the Common Good, was there with 15 other Christian philosophers, theologians and leaders to discuss the implications of AI for society today — and how it can be taught to behave ethically and morally using religion as a guide.

I spoke with Sullivan this week about that meeting. “I’m very grateful for Anthropic’s leadership in this area with faith communities,” she said, noting that most AI companies are not doing that. “It should have happened sooner, but better late than never.”

She was especially glad for the meeting since many people who work in AI seem to think only they matter when it comes to creating this new tool.

“Morally speaking, that has never been true,” Sullivan said. “Religion teaches that everyone matters.”

That was the main idea she and the others emphasized at the meeting: human dignity has to be the bedrock of AI, as it should be for any new technology.

“Christian teaching about human dignity is paramount” when it comes to AI, she said. “That’s what makes life unique. And it is never more relevant than right now.”

The approach she takes when it comes to AI goes by the acronym DELTA.

For her, the “D” stands for the dignity of human beings, putting human life at the centre of AI. “Every human life has value,” she said. “That is a cornerstone of Christianity.”

“E” for embodiment — that we “live in bodies and depend on other people for life,” she said. “We live in time and space, we are vulnerable.”

“L” is for love, which she said contrasts with the dominant ethic in Silicon Valley of optimization — a utilitarian way of viewing the world, as compared to the religious ideal of seeking the best for others.

“T,” she said, is for transcendence, that there is something beyond us as human beings, including truth and beauty.

Finally, “A” is for agency. “Human beings have agency, we can make decisions,” Sullivan said. “AI is trying to supplant that by planning your vacation, your weekly schedule, paying your bills and reading your kid’s bedtime story. We’ve never had the possibility of giving up so much agency before.”

This is especially true when it comes to war.

“We need to talk about parceling out to machines decisions about who and when to kill,” she said, noting it was this issue that got Anthropic in trouble in the U.S. when it refused to let the American military use its AI for fully autonomous weapons.

Some might say the Bible should just be plugged into AI so it can develop its ethical stances. But Sullivan said that would be a bad idea; the Bible doesn’t just talk about things like the Golden Rule and loving neighbours — it also has verses about rape, slavery, violence and the destruction of entire societies.

“That’s the role of religion, of people coming together in community to interpret what the Bible means,” she said. “The answers to those questions will look different for each generation.”

While Anthropic only invited Christians to that meeting in March, it is planning to consult with other faith groups, too.

That’s good news to Sullivan. “Faith groups don’t have an option to sit this one out,” she said of the need for religious people to weigh in. “They need to advocate out of their traditions for human dignity.”

Reflecting on the meeting, Sullivan said she is glad to see a company like Anthropic “take its moral responsibilities seriously … the time spent was a genuine engagement, and not a weak focus group. It didn’t feel like the answers were predetermined.”

Myron Penner teaches philosophy at Trinity Western University in B.C. For him, training AI to be moral is like how parents train their children how to be good people — how to act, what to say, how to treat others.

“My read is that Anthropic is trying to be intentional about introducing moral categories and concepts into AI so that it can function more adaptively,” he said. But, he added, nothing is guaranteed.

“With AI, the “parents” (i.e. programmers) have much more direct control on outputs. But, as also with parents and children, sometimes kids are just going to do what they do. That’s also true with generative AI, which will train itself and come up with responses to particular scenarios that were not programmed.”

Both Penner and Sullivan are looking forward to when Pope Leo releases his encyclical next week on the care of human dignity in the era of AI. “I’m as excited about that as someone who is waiting for the next Taylor Swift album to drop,” Sullivan said.

faith@freepress.mb.ca

The Free Press is committed to covering faith in Manitoba through our Religion in the News project. This reporting continues because readers like you step forward to fund it.

Donate now to support our reporting on religion.
Your donation is eligible for a charitable tax receipt. BECOME A FAITH JOURNALISM SUPPORTER