The Digital Twin and the Line in the Sand

From radio priests to robot dogs: Christian entrepreneurs have always rapidly mastered new technology. AI is no different.

California pastor Justin Lester leads a church (gift link) whose members thought technology was “the mark of the beast” when he arrived, but where today the doors use facial recognition, a robot dog talks to children, artificial intelligence designs Bible studies, and now Lester’s “digital twin” provides online and voice counseling services 24 hours a day. “A lot of our churches are turning into beautiful museums,” he says. “We’re living in the future, whether we like it or not. We can’t afford to sit this one out.”

An evangelical media company called the Gospel Coalition is taking the opposite approach, doubling down on human mediation in a world that’s heading towards AI saturation. They believe it will give them a competitive edge, saying, “What will be worth human readers’ time will be the distinctly human words protected from the dead-eyed, neutering effects of AI.”

Justin Lester and TGC show two emerging Christian approaches to artificial intelligence. These represent middle-class and upper-middle-class responses respectively, one actively embracing, the other techno-skeptical.

I divide new technology into three main layers. The creation layer pioneers the underlying technology, such as neural networks for AI. The platform layer turns the technology into a viable platform that can be built on by others, such as the personal computer or the AI model companies like OpenAI or Anthropic. The application layer builds on platforms to deliver end user value, such as a store’s e-commerce shop or my publication here on Substack’s platform.

With some limited and notable exceptions, American Christians are largely absent from the creation and platform layers. But where they have excelled is in the application layer of actually using technologies and platforms. Every new technology seems to come with a set of religious entrepreneurs who are able to master it and build major institutions or brands off of it.

Father Charles Coughlin built an audience of tens of millions of radio listeners. So did Billy Graham with his “Hour of Decision” broadcasts. Televangelist figures like Pat Robertson became major national figures thanks to their mastery of TV. Earlier, in the 1950s, Archbishop Fulton Sheen won an Emmy award for his TV program. Mark Driscoll built one of America’s largest and fastest growing churches in Seattle in part through becoming an early effective podcaster and user of web 1.0 technology. As you can see from these examples, it’s a mixed record in terms of quality. But Christians were certainly in the game.

The Aggressive AI Adopters

We should expect AI to be no different than previous technologies when it comes to Christians effectively leveraging it. Indeed, we are already seeing pastors leaning into applying this technology and what it can do for them, like the above-mentioned Justin Lester of Friendship Baptist Church in Vallejo, California, the subject of a major New York Times profile (again, gift link). Lester has been a modernizer, innovator and aggressive adopter of technology in general.

The focus of this story is how he used AI to build a “digital twin” of himself that members can turn to for counseling 24 hours a day, including during times when Lester himself is not available:

In the last year, Lester and hundreds of others have brought A.I. into its most human form yet by transferring their thoughts into data and building online replicas of themselves. The digital twins can answer in the real person’s voice and apply their thinking on demand for anyone who wants to chat online or call….Lester cloned his voice and trained his digital twin using almost two million words of his sermons, podcasts, books, text messages and emails. For the last several months, it had been publicly available on his website and Facebook page for online chats or phone calls. It had learned to mimic his bent toward optimism, his obsession with the Green Bay Packers and his casual slang, but it did more than just regurgitate what he had already said. It mapped his thinking onto new conversations to provide advice around the clock for people he’d never met who had questions he’d never been asked.

I’m not going to lie: the idea of creating a digital twin creeps me out. But his parishioners are using it. And he’s helping other pastors create digital twins of their own. This one only seems set to spread more.

Lester is from the black church tradition, which is stereotyped as very traditional, something the article hints at. But here we see the black church on the forefront of sophisticated technology adoption.

His approach to technology also seems consonant with the evangelical group I’ve classified as the “seeker sensitives.” This group is basically the megachurches, often serving middle- or even working-class audiences in the suburbs. Their pastors, like Lester, tend to be highly entrepreneurial, and attuned to cultural trends. Hence, they will likely be quick to experiment with AI. While they may not jump straight to something like a digital twin immediately, once that idea becomes more socially validated, I could see it really taking off. In any case, this type of pastor seems most likely to view AI as a potentially useful tool for spreading the gospel and their religious work generally.

The Techno-Skeptics

Not everybody in the Christian world is so bullish on AI. There’s a strong contingent who are AI skeptical, and somewhat techno-skeptical generally. This group is well-illustrated by the recent AI policy pronouncement from the Gospel Coalition (TGC). Written by Brett McCracken, it is titled, “Our Line in the Sand on AI: TGC’s Missional Opportunity.”

The piece lists TGC’s guidelines for use in writing that appears on their site. Writers are not to use AI for “ideating, researching beyond a basic fact search, organizing, writing, or editing beyond a basic spelling check.” This is one of the stricter AI policies out there. But in light of the various scandals around AI text that have afflicted even big-name publications, it’s entirely reasonable for media outlets like TGC to set pretty tight guardrails for now.

What’s notable about this announcement is not the actual policy, but the rationale. It starts off with an extreme denigration of AI writing:

It’s an offense to the word “writing” to apply it to AI-arranged words and sentences. What AI produces is an impressively polished, usually helpful, sometimes cheeky string of words. But it’s just instantaneous pattern recognition and data organization. It’s soulless slop that can mimic and masquerade as brilliant writing. It’s aggressively efficient, speedily rendered, sycophantic information. But it’s not writing.

Their AI guidelines explicitly say that they have “a low view of AI.”

They also spiritualize their anti-AI stance. McCracken writes, “For Christians, writing is a holy calling and an act of worship (worship is something AI cannot do).” And, “Christ’s incarnation is instructive here. God didn’t reveal himself in a conceptual, virtual, zeros-and-ones sort of way. He took on flesh. Embodiment matters.” And, “Our stance isn’t just strategic for TGC. It’s crucial for the broader Christian mission.” TGC editor-in-chief Collin Hansen posted that he’s making this decision “out of theological conviction.”

I am extremely skeptical of their spiritualized claims here with regards to writing. From a wisdom perspective, I also disagree with their overly negative and narrow view of the uses of AI in the writing process. Where their approach has much more to commend is when it comes to things like pastoral counseling. Their incarnational arguments would be much stronger if directed against something like Lester’s digital twin disintermediating the human relationship with a pastor. The Apostle Paul’s “You know what manner of men we proved to be among you” is what the digital can’t provide.

The important point for us to note here is the strongly AI-skeptical stance being taken. This general posture of techno-skepticism seems to be the default position of the evangelical group that I’ve labeled the “cultural engagers.” These are highly educated evangelicals in the professional-managerial class who are often concentrated in the center of big cities, as well as evangelical colleges and media outlets.

The techno-skeptical stance also seems to be the default in the Roman Catholic world, both in the church hierarchy, and also their lay intellectual classes that parallel the evangelical cultural engagers.

Exploring the AI Future

To summarize, when it comes to AI, and tech generally today, American Christians are mostly absent from the creation and platform layers. They have historically been excellent at applying technology. When it comes to AI, what we might think of as the middle classes of Christian leadership are aggressively adopting it, whereas the upper middle class is very skeptical, and sometimes very opposed to AI adoption.

I’ll caveat this by saying that these are general observations. In a fast moving world like AI, it’s difficult to get stable hard data on this. But there is a Barna survey that finds pastors are rapidly adopting AI. About 50% of them are using it for ideation. A third are using it for research, graphic design, small group questions or administrative tasks. (It’s not necessarily the same people for each of those). Notably, almost a quarter of pastors are using it to write sermons, and only 13% aren’t using it at all. Pastors are more skeptical than the flock about AI - but they are using it anyway.

Again, let’s not over-extrapolate from examples. But I do think Justin Lester and TGC represent two different approaches to AI that are emerging out there, even if the class and group alignments aren’t perfectly clean. If half of all pastors are already using AI for tasks that are prohibited by TGC’s editorial policies, this suggests the approach of more aggressive adoption is winning out. This is what we would expect if seeker sensitives were AI embracers whereas cultural engagers were AI skeptics, since there are far more of the former than the latter.

What does the future look like in terms of how Christians, and Americans generally, will use AI?

My professional history is as a technologist. I spent 20 years developing custom technology solutions (i.e., in the application layer). I was a very good software developer. I’ve solved previously unsolved technical problems, for example, such as figuring out how to automatically recover good data out of corrupted gzip backup files. That one even saved someone’s files at NASA’s Jet Propulsion Laboratory.

I understand technology, am comfortable with it, and am not afraid of it, even if I recognize its potential to disrupt my own work. In general, I’m very bullish on AI. The reality is that going back a long time, many new technologies have created a moral panic, including something as innocuous as the bicycle. We should be aware of that, and avoid allowing ourselves to fall prey to one with AI.

But you’d have to be a fool not to recognize the potential threats AI poses. I am most troubled by the pending intersection of AI with transhumanist ideologies. Far sooner than we think, we’ll be faced with the challenge of cyberization, or the direct integration of technology with human biology and consciousness. Elon Musk is clearly a transhumanist, for example, and it’s no accident he’s both running an AI company and a company trying to build a direct brain-computer interface. It is this, not the almost quaint idea of AI doing our writing for us, that is the real threat to our humanity. This is where the absence of Christianity and Christian thought from the creation and platform layers of technology finds its real life consequences.

But as with the existential risk posed by nuclear weapons, where we were not able to uninvent the atomic bomb, but had to learn to live with it, we are not going to uninvent AI either. We will have to navigate the world with it.

Most likely, we as a society will spring forward until we run into significant problems that cause us to checkpoint, retrench a bit, and move forward more cautiously - but without going back. This seems to be our pattern.

We rushed headlong into large scale industrialization during the late 19th century, then had to tame the beast to make it work for us. This included everything from safety regulations to antitrust to the labor rights movements.

This also happened with no-fault divorce. A heedless Boomer rush into divorce produced the Gen X latch-key kids. Some of the more intentional parenting of Millennial children was a response to this gap. But notably, we didn’t get rid of easy divorce.

It’s similar with social media and smart phones. We’ve seen some of the problems they’ve caused, especially for minors. Now they are being banned from schools, and more limits on them are being imposed. But we aren’t getting rid of them.

Some sort of arc like this seems likely with AI too. It would be nice to think that we’ll move forward into an AI future with everything “aligned,” as the AI safety people put it, in advance. That doesn’t seem likely. Most likely we’ll hit a lot of bumps along the way, but will ultimately find ourselves in a workable equilibrium from which few of us would actually want to go back to the pre-AI world even if we could.

Appendix: How I Use AI

Given my largely pro-AI stance, people have been asking, or even demanding, to know how I use AI in my own writing. I’m happy to tell you. Although I have not yet done it, I would have no problem with using AI to draft functional documents for me - job descriptions, RFPs, contracts, technical documentation, etc. A lot of writing is purely utilitarian, and there’s no reason not to use AI for it.

I do not use AI to write the words in my pieces here or for other sites. As someone who is a rule-follower who even drives the speed limit, I fully comply with any AI policies of any sites I contribute to in my writing for them.

For my own writing here, I do use AI extensively in developing it. I use it to help research and pressure test my ideas. I use it to provide editorial feedback on my drafts. I use it to anticipate objections or subject my arguments to critique. I use it for fact checking, as well as spell and grammar checking. I use it to develop my article titles (note: traditionally, authors never selected their own headlines, and still don’t in almost all publications). It helps me with art direction, to pick a good cover image for some of my posts. I’ve used it on occasion to generate those images, such as for this post. I use it to take my power point mockups and create more attractive designs. I use it to create audio voiceovers of my posts. The one piece of writing it does for me is to draft podcast show notes, largely because if I feed it a transcript, it can automate the otherwise time-consuming task of coming up with a good chapterization.

If I worked for a major publication like the Atlantic, I would have editors, art directors, graphic designers, fact checkers, podcast producers, and maybe even research assistants to help me with my work. As an independent writer without the financial resources to hire people for those tasks, and without the personal skills in some of them, AI is incredibly empowering in helping to close those gaps.

Cover illustration: AI generated.