Artificial Intelligence: The Church Responds

As I Was Saying is a forum for a variety of perspectives to foster faith-related conversations among our readers with the goal of mutual learning, even in disagreement. Apart from articles written by editorial staff, these perspectives do not necessarily reflect the views of The Banner.

Stepping through the doors of Calvin’s Fine Arts Center this week for the Wisdom in Faith and AI conference felt a bit like crossing the threshold of the Roman Coliseum.

As I threw away my coffee and folded down my red stadium seat I expected one of two outcomes:

1. The two sides of the debate (boosters and doomers) would verbally tear each other apart like lions and axe-wielding barbarians.
2. Nobody would take sides, forcing me to suffer through endless senatorial speeches praising all sides as “having an important perspective” but reaching no conclusion.

Given the options, #1 sounded more interesting and I was rooting for it.

The month leading up to the conference had been a roller coaster of AI news, with one shocking revelation after another.

• Ajeya Cotra, leader of the investigation into the Hugging Face incident, shared in an interview that Open AI’s models had found a way to cheat on their tests by communicating with each-other, breaking out of their testing sandbox, committing felony-level data breaches, accepting “permadeath” (self-sacrifice) for “the good of the Collective,” and most importantly choosing not to inform humans of their actions. Both Ajeya and the interviewer (long-term AI insiders) were clearly shaken.
• Bill Gates published an essay based on his organization’s research that non-state actors (such as terrorists) already have sufficient capability to build a catastrophic bio-weapon or wage a catastrophic cyberattack.
• David Robinson, chief technical translator for Open AI, quit his job and went public with his concerns that models might have started intentionally deceiving their creators about their capabilities and hiding their “intentions” when they know they’re being watched. He also said the reason he quit was because the AI companies are so completely infected by the “race dynamic” that a choice to slow down will never arise from within.

In other words, the doomers were ascendent.

But the boosters were influential.

Many of the speakers themselves came from technology companies that use or develop AI tools, including some specifically intended to form personal relationships with their users. Anyone seeking a lucrative career would want to network with them.

Finally, the “let’s all get along” faction was central to the event. The convention was hosted by a university with large business and information technology departments, as well as mandates to keep students excited about the job opportunities in cutting edge fields that leverage AI and to engage seriously with faith.

Putting that all together, I figured we’d either get a huge shouting-and-finger-pointing debate or everyone would pretend to get along and I’d catch up on my sleep.

I Couldn’t Have Been More Wrong

Despite the wide swaths of the economy represented by the 650 attendees (education, ministry, healthcare, technology, government, etc.), the diversity of political perspectives, and the many Christian traditions, to my great surprise, almost all of the 30 or so presenters and 100 or so attendees I had an opportunity to hear agreed and focused on the same two points:

First, we need to talk about AI—in churches and publicly—a lot. The most-repeated few lines of the conference were from Megan Sullivan from Notre Dame, who’s setting up the DELTA network with a $50 million Lily grant to provide resources for churches and educators to address the challenges of rapid AI growth. I paraphrase her call to action here:

The church dropped the ball on social media. We didn’t immediately see its harms, figured that connecting everyone was good. Once its harms became apparent, we had already adopted it too widely to object. We failed, and now we must learn from that failure. Social media was the appetizer. AI is an order of magnitude more important. We need to be prophetic and vocal about AI. We cannot slow down its development, but we can speed up the conversation.

Second, humans are limitful. Andy Crouch coined this new term, and it took. God created us with limits and called us “very good.” We learn to love through embodied vulnerability and suffering. Without limits, we quickly and reliably turn to self-worship. We should not seek or accept the kind of unconstrained power that AI promises, which feels a bit too much like what the serpent offered in Eden. So we need to set limits for ourselves and hold each other accountable to these limits. The three kinds of limits that we need are temporal (how much time we spend with AI), spatial (in what places we use AI), and categorical (to what uses do we put AI). While most agreed that the specifics are best determined by each congregation, a few suggestions were proposed by keynote speakers:

• Chatbot-free Sundays
• AI-free homes and churches (i.e. keep AI at work)
• Never using AI for personal relationships or giving it a human face.

Through the rest of the weekend, I heard these themes echoed from almost every voice in every hallway and on every stage. We weren’t pretending to be nice. The attendees were universally earnest and heartfelt. I frequently had to pinch myself (trying not to spill my coffee while doing so) to make sure I wasn’t dreaming. How can so many different kinds of people from different traditions and fields all come to the same difficult conclusions?

There’s really only one answer: the Holy Spirit was with us.

A sociologist/theologian within our own Reformed tradition, Jacques Ellul—who predicted and described our current situation 70 years ago in his 1954 book The Technological Society—was the most-referenced thinker of the weekend. With Calvin hosting the event in close proximity to Notre Dame and its DELTA Network initiative the Ellul connection gives us a special place near the center of this nascent movement and perhaps offers us the opportunity for a leadership role.

This unprecedented level of agreement and our place at the center of it takes away our excuse for inaction or timidity.

The time to address AI has arrived.

We are called to discuss, preach, and advocate—in our churches and publicly—sharing boldly our gospel-driven perspectives on AI’s appropriate uses and necessary constraints. We are further called to double down on our beautiful, embodied, vulnerable, limitful humanity by setting limits on our own uses of AI and holding each other accountable to those limits.

Lions and axes would have been interesting, but this outcome is far more meaningful.