Think back to the last hour you spent behind the wheel. Or the last time you pushed a mower around the yard, or stood at the sink working through a stack of dishes. Your hands were busy, but the road, the grass, or the dishwater asked almost nothing of your mind. So your mind was free...free to wander, to process, to think.

We tend to call that dead time. But I have come to think of it as some of the most alive reading time I enjoy.

I want to name this tension out loud, because I suspect you feel it too. Ministry leaders are told, constantly, that we should read more. We have to read theology to keep our roots deep, read leadership books so we can lead well. We need to read widely, for continuing education and for the sake of our own souls. We may nod in agreement, because we know it’s true. But then we go home to lives that have no chair-time left in them. And so the reading we know we need (or want) to do sits on a nightstand, judging us, while the day eats every quiet minute we might have given it.

But what if the problem was never time? What if instead it’s the format?

The Lifeline, Not the Compromise

I have now spent forty years in ministry, most of it bivocational (or “co-vocational” as we call it now). I know what it is to come home with the tank empty, having already spent the day’s focus somewhere else, and to face that stack of books with no energy left to give to it. For leaders living that life, the advice to “just read more” can feel less like encouragement and more like one more way we’re falling short.

So let me say plainly what I believe, because I believe it strongly: audiobooks are not a lesser substitute for real reading. For a leader with no margin, they are often the only way reading happens at all. If that sounds like I’m lowering the bar, consider that the man who set the bar impossibly high for generations of ministers would have absolutely agreed with me.

John Wesley was relentless about reading. When a traveling preacher named John Trembath excused his thin, repetitive preaching by admitting he simply had no taste for books, Wesley refused the excuse. In a letter dated August 17, 1760, he told him:

That certainly wasn’t the only time he gave such advice. In Minutes from the conference, he said, “Read the most useful books” (Leeds, 1766). He famously advised his traveling preachers, “Steadily spend all the morning in this employ, or at least five hours in twenty-four.” And when he heard excuses (like Trembath’s), he closed the door on every excuse against reading with a line that still stings:

“Contract a taste for it by use, or return to your trade.” In other words, read, or go back to your old job. NOTE: There is a quiet irony I can’t resist for a bivocational audience: Wesley told a preacher to read or go back to his trade, never imagining a generation of us who would do both at once.

But John Wesley held himself to that exacting standard most of all. To make those hours of reading and study possible amidst a brutal itinerant schedule, he famously read on horseback, propping his books against the saddle as he covered thousands of miles. He even had a special travel desk constructed to adapt his saddlehorn for reading. Wesley refused to let a life of itinerant ministry cost him his depth. So he found his margins in the saddle.

That is reading in the margins, three centuries early. And I am convinced that a man who read on a moving horse would not have hesitated for a second to read with his ears while his hands and the road were busy. We have far better margins than a saddle. We have the commute, the mower, the sink, and the treadmill. A leader who has run out of chair-time can now meet a standard that print alone makes impossible.

One caution, though, and Wesley would insist on it. Even he, who pushed reading this hard, warned against making an idol of it:

Reading in the margins is meant to feed the soul and the ministry, not to turn every spare second into one more thing to optimize. Hold that thought. We’ll come back to it.

What Changed for Me: What the Numbers Show

I have been an Audible member since 2017. That’s nine years now, long enough that this is not a reaction to a shiny new app or a current sale. It’s a near-decade-long practice, and it changed reading for me from an aspiration into a habit.

I’m also a data and analytics person, so let me show you rather than just tell you. I have always loved what analytics reveal, because numbers surface truths about our own behavior that we simply cannot see in the moment, when we’re too close and too busy to notice the pattern.

For 2025, my Audible reading stats said I explored 21 titles and finished 12 of them, averaging 38 minutes a day on the days I listened. Audible even handed me a “Genre Hopper” badge, which truly fit, because those books I read last year ranged across artificial intelligence, future tech and innovation, leadership, theology and ministry, teaching, fiction, and even a little quantum physics.

But that wasn’t the only such highlight. I haven’t shared my stats every year, but I did see in what I shared in 2021 that the pattern still holds: in a single year I once finished more than thirty books across audio and print combined, and yet I was only averaging around fifteen minutes a day listening. Even at that gentle pace, the margins compounded into reading roughly a book every week and a half!

That’s the part that still surprises people. There aren’t many people today who can find an extra five hours a week to sit down and read uninterrupted. But almost everybody has a commute, a yard, a sink, a drive, or some downtime before the next meeting. Same for me: the minutes were always there. Audio is simply what finally let me pick them up.

But is Listening Really “Reading?”

I know the objection, because I’ve heard it many times, and perhaps you’ve thought it: listening isn’t really reading – it’s easier, it’s passive, so it’s somehow second-class.

Here’s why I think that instinct, however common, has history exactly backward.

For most of the Church’s life, believers did not read the Word. They heard it. Literacy was rare and books were scarce, so Scripture was overwhelmingly an aural experience, something proclaimed aloud to a listening community. That’s not a workaround the early church settled for. It’s the native mode of the faith.

“Faith comes from hearing,” Paul wrote, “and hearing by the word of Christ” (Romans 10:17, NASB). He told Timothy to “devote yourself to the public reading of Scripture” (1 Timothy 4:13, NIV), which meant reading aloud to people who would receive it with their ears. At the inauguration of Jesus’ ministry, in the synagogue in Nazareth, Jesus read aloud from the Torah scroll from Isaiah, as was his custom:

Whether it was the lector reading to the gathered assembly, the monastic community listening silently as the text was spoken over them, or Zinzendorf’s Moravian disciples listening reverently to the Word of God in monastic reflection, this is the deep tradition. Silent, private, eyes-on-a-page reading is the historical newcomer, not the audiobook. After all, the printing press didn’t come along until about 1440, so that’s well over 2,000 years of church history with no printed books, only scrolls and carefully hand-calligraphed rare copies chained to pulpits.

This isn’t an abstract point for me. One of my favorite discoveries in my doctoral research was Lucy Lind Hogan’s chapter on the Sermon on the Mount as live spoken discourse (in Preaching the Sermon on the Mount: The World It Imagines). Drawing on her own experimentation with delivering the Sermon as oration rather than reading it on a page, she raised questions about how that text lands when it is heard aloud as it was first delivered, questions that genuinely reshaped how I think:

Given my dissertation’s focus on discourse analysis and narrative, it was a meaningful insight: a text spoken and heard is doing something a text silently scanned is not. We did not lose something when we moved from hearing to reading. We simply traded one set of strengths for another. For all of its convenience, audio quietly hands some of the first strenghts of orality back.

So no, I don’t think listening is a lesser way to take in a book. In many ways, it’s the oldest way the faith ever read Scripture.

What I Actually Listen to

Let me get practical and tell you what’s actually been in my ears, because a recommendation is only as good as the specifics behind it. I’ll group these the way I actually use them.

Books that have shaped my thinking. My favorite here is Geoff Woods’s The AI-Driven Leader. For anyone trying to figure out how to navigate AI adoption and integration in an organization, whether a church, a business, or a nonprofit, this book is full of genuine insight and actionable advice. I have recommended it often, and I’ll recommend it again right here.

Alongside it, I would include:

• Mark Coeckelbergh’s AI Ethics,

• Diamandis and Kotler’s The Future Is Faster Than You Think,

• Pascal Bornet’s Agentic Artificial Intelligence, and

• Urs Gasser and Viktor Mayer-Schönberger’s Guardrails,

Books for ministry leadership. My favorite recent read among these is Tod Bolsinger’s Canoeing the Mountains, and I keep returning to it. Its premise, leading well when the terrain no longer reflects the map, has shaped my own work more than almost any other leadership book. I drew on it in the Conclusion of Early Warning, I built on it in a talk I gave for pastors and ministry leaders on navigating technology disruption in ministry, and I returned to it again in a recent podcast interview with Chris Conrad about adaptive leadership. If that theme grabs you, hold onto it, because it’s where I’m headed soon here on τέκτων.

Other audiobooks I’d highly recommend from this shelf include:

• Mark Batterson’s In a Pit with a Lion on a Snowy Day

• Gordon MacDonald’s Who Stole My Church?

Fiction for restoration. And yes, I mean restoration, not just entertainment. Remember John Wesley’s warning about being swallowed up in books? I do read a significant amount of non-fiction, history, theology, and technology, but periodically enjoying fiction in the margins is part of how I keep reading from becoming one more performance metric. My favorite audiobook from the last year here is Lucas Kitchen’s Missionary to Mars, and there’s a personal reason it delighted me. Years ago, I told a room full of ministry leaders that they needed to get ready, because before too long, we would be sending missionaries to the colony on Mars, and their own grandchildren could very well be the ones who rise to that call. Then I came across Kitchen’s novel, which fleshes that very idea out beautifully...and he had never heard a word of what I’d been saying. That kind of recognition is its own quiet joy.

For more Christian fiction for restoration, I would also recommend:

• Tracy Higley’s A Time to Seek, A Time to Weep, and A Time to Love - a 1920’s time travel archaeology adventure mystery series

• Madeleine L’Engle’s A Wind in the Door - a powerful illustration of spiritual warfare and the power of relentless love over hate and evil

The Narration Question (and a Question for YOU)

This is where I want to slow down, because as a τέκτων, a craftsman thinking about his tools, the matter of who reads the book to you turns out to matter a great deal. I do have some standing to care about this. My own daughter trained professionally as an audiobook narrator and has voiced more than a dozen titles on Amazon. I have watched up close how much craft goes into reading a book well.

So here’s my honest assessment: At its best, an author reading their own work is a delight. I loved listening to Lucas Kitchen narrate Missionary to Mars himself (voices and all), and I’m thoroughly enjoying Anthony Hunt reading his own book on the AI-powered church right now, because his enthusiasm for the subject is unmistakable in a way no other voice could imitate. But “author-read” is only wonderful when the author can actually read well. If an author reads flatly or with distracting quirks, a trained professional narrator will outshine that every time.

And then there’s AI narration, which is now expanding the catalog dramatically for better or for worse. You’ll notice those Tracy Higley titles I mentioned were generated by “Virtual Voice,” Amazon’s AI narration. I’ll be candid. They were mostly good, and most of the time it was quite convincing and realistic, but there were some very clear (& sometimes frustrating) tells: unnatural pauses, the same Egyptian character’s name pronounced several different ways, and lines that should build, express surprise, or reveal slowly read out uncannily flat instead. But, there is an honest tradeoff, too: I would never have made time to read several of those titles in print at all. The convenience was worth tolerating an imperfect voice.

I also know that there are some powerful new voice generation tools available through ElevenLabs, and my favorite video translation tool, HeyGen. So I think what we see emerge in the future will become mind-blowingly realistic. We’re just not quite there…yet, at least for audiobooks. My one firm conviction on this is that AI narration should always be disclosed, so the listener knows what they will be hearing and can decide what they want to listen to.

I’m wrestling with all of this personally right now, because I hope to release an audiobook edition of Early Warning soon:

• Do I record it myself?

• Contract with a professional narrator?

• Turn to AI?

Each approach has pros and cons, benefits and risks, and I genuinely haven’t landed. However, that’s a decision I’ll be making in the next few weeks.

So I’ll turn it over to you, because you’re my potential future listeners. Which do you prefer when you hit play? The author’s own voice, a trained professonal narrator, or a well-made AI reading? If you’ve got an opinion on which Early Warning should be, I’m all ears. Tell me in the comments.

For Other Ministry Leaders

I’ve written this mostly from where I sit, but I think the case for audio reading extends across nearly every role in ministry.

• The worship leader can pull devotional and craft input into the cracks between rehearsals and sound checks.

• The communications director can stay culturally current on the drive in, instead of falling behind the very conversations they’re called to engage.

• The professor or chaplain can turn a commute into continuing education, one chapter at a time.

• The board member or lay leader can finally get through that leadership title before the retreat, rather than carrying it, unread, into the meeting.

Different roles, same principle. The margins are everywhere. Audio is what lets us pick them up.

Two Entryways: How to Choose

If this article has you ready to give Audible a try, let me save you some confusion, because Audible’s plans can be a bit confusing at times. There are really two different philosophies of how you acquire books here, and the right one depends entirely on which kind of reader you are.

1. The ACCESS model (Audible Standard). Think of it like a streaming subscription for your ears. You pick a title each month and listen, but the books live in your library only as long as your membership is active. With this model, you’re paying for access, so when you cancel, they’re gone. This is ideal for the reader who consumes books in short windows…someone who will happily devour a book the month they choose it and doesn’t need it on a permanent shelf afterward. For simply testing whether listening fits your life, it’s a near-perfect, low-cost on-ramp. Standard runs $8.99 a month: https://amzn.to/4eq6mEa.

2. The OWNERSHIP model (Audible Premium Plus). This works like buying books, one credit at a time (since books vary in price, a “credit” is the currency, as one credit equals one book). Your monthly credit buys a title that is yours to keep forever, even if you cancel. Over time, that builds into a permanent, reusable reference library. This is the model for the reader who returns to books: the one who will pull a leadership title back off the shelf for a board retreat two years later, or re-listen to a chapter while prepping a talk. That’s how I use it, and I now have 159 audiobooks in my app library. Premium Plus is $14.95 a month, or $149.50 a year (& gives you one credit each month): https://amzn.to/3S4K5Es.

(There’s also a bare-bones Audible Plus tier at $7.95 a month, but it’s streaming-only with no credits, so it’s not really built for keeping books.)

You can also gift an Audible membership to others (1, 3, or 6 months, or even a full year). And if you don’t have Amazon Prime but want to get in on the Prime Day deals, they’re running a special right now (and an awesome option for young adults): https://tinyurl.com/AmazonPrime-YoungAdults.

Now the catch nobody tells you, and I’d rather you hear it from me. If you sign up specifically to grab a ministry or AI title you’ll want to keep and reference for years, know that on the free Standard trial, that title will disappear when/if you cancel. To own it, you want the Premium Plus path, even if that just means taking the Premium Plus trial for one keep-forever credit. One more honest note from long experience: those keep-forever credits are wonderful, but they can quietly pile up unused, and an eager reader can outpace their credits and start buying extra titles (there is an option to scale back to other plans, including one that’s lower cost and only offers one credit every other month, but you can’t find that online or through the app; usually have to call Audible’s Customer Service to request that).

Here’s the timely part, and the reason I’m writing this now. Audible regularly runs trials, but at the moment there’s an especially strong entry point tied to Prime Day. Through July 15, Prime members can get three free months of Audible Standard, normally $8.99 a month. If you’re not a Prime member, there’s a parallel offer of three months of Standard for $0.99 a month. And Prime Day itself runs June 23 to 26 this year, which is exactly when a lot of us (my wife included) like to line up our bigger or subscription purchases to stretch the savings. If you’ve been meaning to try this, that window is a genuinely good moment to start. Just remember the deadline for the three free months deal is July 15.

What it’s Not

Let me be as honest about audio’s limits as I am about its strengths, because choosing the right tool means knowing when a tool is wrong for you.

Audio is the wrong choice for some kinds of reading, and I won’t pretend otherwise. Dense study texts you need to mark up, books that are thick with charts, visuals, and footnotes, and close exegetical work on a passage all belong in print or on a screen where you can annotate, flip back, and sit with the page. This connects to something I wrote here not long ago in “After, Not Instead Of.” Just as I argued there that AI belongs after the preacher’s own hard work rather than in place of it, audio belongs alongside deep study, not instead of it. Listening is superb for breadth, exposure, and formation. It feeds the well, the illustrations, the cultural awareness, and the theological insights you draw on for months. But the close exegetical work at the study desk is still its own discipline, and audio doesn’t replace that.

And there’s one more thing about audio that the “convenience” factor undersells. For many of the leaders I’m writing to, audio isn’t merely the convenient way to read. It’s the only way that fits the life they actually live. The bivocational minister working a trade, driving a route, doing manual labor with hands full and ears free. The leader who reads slowly, or lives with a vision difference, dyslexia, neurodivergence, or a processing difference that makes a printed page a wall. For these folks, “lifeline” isn’t a figure of speech. It’s the literal truth. Audio doesn’t simply help them read more; it’s what enables them to read at all.

What You Can Do Next

So here’s where I’ll leave it. A tektōn chooses tools that fit the work. For the leader who has run out of quiet chair-reading time, the right tool was never more hours in the day, because those aren’t coming. It’s a different format, one that turns the margins you already have into a library.

Think back to that drive, that mower, that sink full of dishes. The time was never dead. It was waiting. Wesley found his reading on the back of a horse. We can find ours in the ordinary, hands-busy, mind-free moments that fill an honest week of ministry. Faith comes by hearing. So, it turns out, can a great many other good and worthy things.

* As I share in my About page, I have affiliate links enabled for the products I’ve identified & linked above, but this is NOT sponsored content. Every recommendation here is a personal one, for tools, products, and titles I genuinely use and strongly believe in.

Resources referenced in this article:

Bolsinger, Tod. Canoeing the Mountains: Christian Leadership in Uncharted Territory. Downers Grove, IL: InterVarsity Press, 2015.

Lucy Lind Hogan, “You Be the Judge (Matthew 7:1-6),” in Fleer, David & Dave Bland (eds), Preaching the Sermon on the Mount: The World It Imagines (Atlanta, GA: Chalice Press, 2007), 152.

Swisher, David J. “After, Not Instead Of.” τέκτων (tektōn). June 8, 2026. https://tektonministry.substack.com/p/after-not-instead-of

Swisher, David J. (2017). “Vantage Point: Using Narrative and Discourse Analysis of Jesus’ Sermon on the Mount to Improve Discipleship Outcomes.” Portland Seminary at George Fox University. Doctor of Ministry, 238. https://digitalcommons.georgefox.edu/dmin/238

Wesley, John. “Chapter XII – In Conference with the Preachers.” In John Wesley the Methodist. Wesley Center Online. https://wesley.nnu.edu/john-wesley/john-wesley-the-methodist/chapter-xiii-in-conference-with-the-preachers/

Wesley, John. “Letter to Joseph Benson, WYCOMBE, November 17, 1768.” In The Letters of the Rev. John Wesley. Wesley Center Online. https://wesley.nnu.edu/john-wesley/the-letters-of-john-wesley/wesleys-letters-1768/

Wesley, John. “Letter to John Trembath, CORK, August 17, 1760.” In The Letters of the Rev. John Wesley. Wesley Center Online. https://wesley.nnu.edu/john-wesley/the-letters-of-john-wesley/wesleys-letters-1760/

There are a lot of people writing about AI and the church right now. Not many of them have stood at the front of a real congregation on a Sunday morning and wondered how to lead people faithfully through a disruption they did not choose. Even fewer have done graduate-level research on AI’s impact on the Church. David Betts has. That matters more than you might think.

Betts is an an Associate Pastor at Deer Park Alliance Church, and an academic advisor at Okanagan Bible College. He was previously the lead elder at Trinity Church in Red Deer, AB (now called Centre Church). He also devoted a significant chunk of his Master of Divinity research to exploring the relationship between church leadership and artificial intelligence. The Church and AI: Seven Guidelines for Ministry on the Digital Frontiers grew directly from that academic work. At 270 pages and $7.36 (U.S.) in paperback, it is one of the most accessible and affordable books on AI for ministry.

Research from the Trenches

There are a lot of ways to write a book about AI and the church. You can approach it theoretically, working through the philosophy and theology from a comfortable academic distance. You can approach it journalistically, cataloging what is happening and pointing out trends. But David Betts does something different: he brings pastoral experience and serious academic preparation to the same table, and that combination gives the book a grounded quality that not many resources in this space can claim.

His MDiv research on AI and ministry shaped the organizing framework of the book, which moves through AI’s history and development, its potential impact on society, its theological implications for the church, and then into his seven guidelines for church leadership. That sequence matters. Betts does not start with the guidelines. He earns them.

What is worth noting is that the foundational essay behind this book is publicly available. One of his earlier posts on his Substack, Church and AI, was the academic paper he submitted as part of his MDiv work. If you want to understand how he thinks before you invest in the book, that is a good place to start:

Church and AI

Seven Guidelines for Church Leadership in an Increasingly AI-driven World

Author’s note: this is an academic essay submitted for a Masters of Divinity module. Be warned: it’s about the length of a book chapter, so it’s not for the fainthearted! In the near future, I’ll be looking to present snippets from this paper in a way that is more casual for the general reader. My prayer is that the content presented in the study will s…

Read more

3 years ago · 1 like · 1 comment · Dave Betts

The Seven Guidelines

The heart of the book is the seven guidelines that Betts proposes for church leaders. Here they are, with brief context drawn from his research. One thing worth noting before you read them: Betts runs an extended nautical analogy throughout Part Two. Your crew represents the priority of relationships; your ship represents a resilient congregation; your sails represent adaptable structures; your tools represent embracing technology wisely. It is a memorable device, and it gives the section a cohesive, accessible feel.

1. Prioritize relationships: AI will enhance administrative capacity and transform many ministry tasks, but it cannot replicate authentic human connection. Betts argues that pastoral presence and genuine community will become more essential, not less, as AI proliferates.

2. Nurture resilient congregations: Congregations need to be equipped theologically and spiritually before the disruptions arrive. Building biblical literacy and spiritual depth now is proactive preparation, not just good discipleship practice.

3. Build adaptability into church structures: Ministry contexts will keep changing. Churches that build flexibility into their staffing, processes, and planning will navigate disruption far better than those that are rigidly structured.

4. Embrace positive technological developments: Betts is not calling for uncritical adoption. He is calling for discernment over reflexive resistance. The question is not whether to engage with AI, but how to do so in ways that serve the mission.

5. Stay informed in a rapidly changing environment: Church leaders do not need to become AI experts. They do need to pay enough attention to pastor their people through the cultural shifts that AI is already creating.

6. Be proactive in praying for God-given wisdom: Betts draws a clear distinction between the information AI can provide and the wisdom that only comes from closeness with God. In a world increasingly oriented toward AI as an answer source, this guideline feels especially urgent.

7. Keep an undistracted focus on the mission of the Church: Whatever changes, the mission does not. Betts anchors this guideline in the sovereignty of God and the certainty of Christ’s return.

That last one deserves a moment. But first, consider the observation Betts uses to open his chapter on wisdom. He leads with a quote from Isaac Asimov:

That is the tension Betts is working with throughout Guideline 6. In a world where AI delivers more information than any generation has ever had access to, he argues that the church’s response cannot simply be to consume more of it. Biblical wisdom is not a data problem. It is a proximity problem, one that requires closeness to God, and that is not something any algorithm can produce. Betts writes with the confidence of someone who has thought carefully about the worst-case scenarios for AI and still believes the story ends well. He anchors Guideline 7 in the sovereignty of God and the certainty of Christ’s return, arguing that whatever awaits the Church in an AI-driven future, the mission is unchanging and, with God, eminently possible.

Shared Perspectives

Also, as you can probably see, Betts and I share many of these same themes. My book Early Warning echoes many of those guidelines, even though I had not read Betts’ work when I wrote mine. Also, I have always loved nautical imagery, and I utilize it as a key metaphor in my forthcoming book Navigating the AI Fog: Perspectives on the Responsible Use of AI in Ministry. So, that’s one of the reasons I excitedly asked David Betts to contribute a chapter.

What Readers Are Finding

The book has earned a 4.4 out of 5 rating from six readers on Amazon (and the only non-5 rating came from someone who left no comments). The substantive reviews reflect a consistent theme: readers appreciate Betts’s refusal to embrace either extreme.

One reviewer described his approach as a call to be “technological semi-Luddites,” meaning neither fearful reactionaries nor uncritical enthusiasts. Another noted that Betts “will not send you running for the hills in fear or surrendering your brain to the whims of AI,” but instead challenges readers to engage with both the possibilities and the risks through the lens of faith. Both reviewers also highlight something worth flagging: the book keeps discipleship at the center. For Betts, AI is never the main character. The mission of the Church is.

It is also worth being clear about what the book is not: Betts states plainly in the introduction that if you are looking for a guide to implementing AI tools in your church, you will likely finish disappointed. This is a book about broader societal, cultural, and theological implications, not a practical how-to guide. That is not a weakness; it is a deliberate choice. But it is worth knowing before you begin. The research behind it is serious, and the pastoral instincts he communicates are sound.

What This Means for Your Ministry

Here is the question I find myself returning to after reading a book like this: where are your people right now?

Your congregation almost certainly includes people who are already using AI tools in their work and personal lives. Some are enthusiastic and perhaps not asking enough questions. Some are worried and not sure why. Some are simply confused. What they all share is a need for pastoral leadership that can help them think Christianly about something they did not choose and cannot avoid.

Betts’s seven guidelines are not just for senior pastors navigating institutional decisions. He wrote this book for church leaders, ministry leaders, and laypeople alike, and the guidelines reflect that breadth. They are as useful for a worship leader or small group director trying to think through AI’s implications as they are for a lead pastor making policy decisions. His emphasis on relationships, for instance, is a reminder that pastoral presence is not a legacy function that AI will eventually replace. It is the irreplaceable center of what you do. His call to nurture resilient congregations means helping people build the spiritual depth now that will hold them steady when the disruptions intensify…and they will.

His call to pray proactively for God-given wisdom is the one that speaks most naturally to a pastoral audience. The invitation is not to be anxious about AI or to ignore it. It is to bring it to the Lord and ask for the kind of wisdom that no language model can generate.

And his insistence on keeping an undistracted focus on the mission is not a way of avoiding hard questions. It is a way of keeping them in their proper place. Ministry leaders can spend enormous energy reacting to AI news cycles and policy debates. Betts’s framework helps redirect that energy toward formation, discernment, and the work that actually matters.

What You Can Do Next

The Church and AI is available in paperback, and definitely worth getting. The price point makes it realistic for everyone on a leadership team to read it together, and a guided discussion through the seven guidelines over several meetings would be a practical way to apply it to your specific context. You can learn more about his approach, and stay up on his latest thoughts and posts through David Betts’s Substack, Church and AI at churchandai.com. His foundational essay on the seven guidelines is publicly available there, free of charge, and it will give you a strong sense of his thinking and research before you commit to the book.

The questions Betts raises are worth sitting with:

• How does AI affect your pastoral care approach?

• What does faithful stewardship of AI actually look like in your setting?

• How do you help your congregation build the resilience it is going to need?

This book can help you start those conversations, and starting them now is wiser than waiting until you have to.

Book Details and Links

Betts, David. The Church and AI: Seven Guidelines for Ministry on the Digital Frontiers. Red Deer, AB: David Betts, 2024. 264 pages. Available in paperback and Kindle (no audiobook edition).

• Order The Church and AI on Amazon #AffiliateLink

• More reviews of The Church and AI on Goodreads

• David Betts’ Substack Church and AI

My Curated List

Want more insights on useful books on AI for church and ministry contexts?

I maintain a curated list on Raindrop.io of all of the available books on AI for pastors and ministry leaders. It is available at https://WildcatTech.raindrop.page/ai-for-church-and-ministry-leaders-65707585. To the best of my knowledge, nobody else has anything comparable.

Resources Referenced in This Article

Betts, David. The Church and AI: Seven Guidelines for Ministry on the Digital Frontiers. Red Deer, AB: David Betts, 2024. https://amzn.to/4qtHtMC

Betts, David. “OUT TOMORROW: The Church and AI - Seven Guidelines for Ministry on the Digital Frontiers.” Church and AI. April 30, 2024. https://www.churchandai.com/p/out-tomorrow-the-church-and-ai-seven

Betts, David. “Seven Guidelines for Church Leadership in an Increasingly AI-driven World: Imperatives for leading well in a rapidly changing church context.” Church and AI. July 31, 2023. https://www.churchandai.com/p/leading-churches-into-an-ai-world

Swisher, David J. Early Warning: How to Lead with Wisdom in Times of Technological Disruption. Plano, TX: Invite Resources, 2024. https://amzn.to/4xiwOYR

Swisher, David J., Editor. Navigating the AI Fog: Perspectives on the Responsible Use of AI in Ministry. Tekton Consulting. Dec. 2026 (forthcoming). https://responsibleuseofai.org/anthology

There’s a barrel organ in a historic church in Cambridge, England whose story I find fascinating, and the lessons I’ve learned from it continue to shape and influence my ministry. I first stood in front of it at Holy Trinity Church (Cambridge, UK), where Charles Simeon installed it in 1793. By the standards of his day, it was a scandalous choice: an automated mechanical instrument, the kind street performers used, brought into worship. Simeon’s critics were certain it had no place there. He was convinced it could reach those not yet reached, and so he consecrated it to the mission. History tells us which of them was right. The instrument is still there, and the revival it helped accompany lasted fifty years. (I tell that story in full in Early Warning, and I explored it here on τέκτων not long ago.)

What has stayed with me is not the organ itself, however. It is the shape of the objections raised against it, because I have seen that pattern of objections so many times since. For the last 2.5 decades, I have worked bivocationally at the intersection of higher education technology and ministry, and in both contexts, I have watched the same pattern of objections repeat with almost every new tool that emerges:

1. It comes from the secular world

2. It replaces something dignified with something popular

3. It will attract the “wrong” kind of people

4. It is mechanical rather than genuine, and

5. We’re not sure where this will lead.

The specific technology changes with every generation. But the pattern of resistance, and the eventual fruit when faithful leaders push through it, does not.

Subscribe now

So when I was asked to develop a workshop on preaching and artificial intelligence for Anderson University’s National Preaching Clinic 2026, hosted by the James Earl Massey Center for Compelling Preaching, I decided to lead with Charles Simeon’s story on purpose. I wanted to put that familiar pattern on the table before I ever said the word, “A.I.”

Then something happened that encouraged me far more than I expected. I heard it…distinctively. The quiet “ahs.” The low “um-hums” rolling across the room. The nods of affirmation before I had finished telling the story. It was clear that many of these preachers recognized the pattern instantly, because they had lived it. I had not touched a nerve so much as named something they already knew in their gut. And that recognition, it turned out, was the doorway into the real conversation.

Developing that workshop took weeks of prayerful research and preparation, days of thinking, praying, and writing, along with a good deal of ideative collaboration with my favorite AI tools (I wanted to practice what I preach). The framework I am about to share with you is the very one I used to prepare my talk, as well as one of the key takeaways I shared in it.

The Arguments Always Sound Familiar

I did not have to reconstruct those objections from history books. I recognized them on sight. Why? Because across more than two decades of working bivocationally in technology innovation roles, I have watched that same script play out in both contexts. It sounds one way in a university faculty meeting when someone proposes a new learning platform, and another way in a church board meeting when someone suggests a livestream camera in the sanctuary. But underneath, it is the same handful of concerns, lobbed against essentially the same desires for innovation and outreach.

And the lineage is long. The church raised versions of these very objections against the printing press, against the microphone, against the overhead projector. Faithful leaders pressed through every one of them, and the mission was better for it. That is the pattern I wanted to name out loud before we turned our attention to the newest tool on the list.

The Real Question Was Never “Whether”

What encouraged me most was not the framework I brought. It was where I found the room already standing in agreement.

Going into my preparation, my rough estimate was that maybe ten to fifteen percent of the room would be deeply skeptical and resistant to AI altogether, and likely another ten to fifteen percent would be eager adopters already using it, and I figured that the broad middle (perhaps seventy to eighty percent), would be somewhere in between. As best I can tell, those numbers held fairly close to accurate. But the range of receptivity was not the surprising part (I’m used to that). The surprising part was where the two ends met…an intriguing convergence.

Even among the resistant, I sensed a quiet agreement that the need was real, and that the live question was not whether to use AI but how and when. And even among the eager adopters, when I shared my cautions, when I talked about how not to use it, when not to use it, and how to recognize when you have become too reliant on it, I saw the same nodding. Both ends of the room, the skeptics and the enthusiasts, had arrived at the same place. The debate was no longer for-or-against. The debate had moved to a better question: what does faithful, responsible use actually look like? I found that tremendously encouraging.

That convergence is the most important thing I can tell you about where this conversation is landing in churches right now. And, fortunately, it is not just my read of one room, either. Recent research from Barna, in partnership with Gloo, found that Christians are extending trust and registering fear about AI in the same breath, holding openness and concern together rather than resolving them. The tension I felt in that room is the same tension the data is measuring across the country.

And this is not only happening in the church. A national study from the National Bureau of Economic Research found that personal use of ChatGPT has overtaken work use, climbing from 53 percent to more than 70 percent of all messages. The everyday questions we once brought to a friend, a mentor, or a pastor are increasingly going to AI first (and OpenAI’s “Signals” data validates these trends, too, with a growing concern about the need to strengthen how it handles mental health conversations).

There is a reason that tension matters so much, and it raises the stakes for everything that follows. The people in our pews are not waiting for us to make up our minds. Barna found that a meaningful share of Christians are already prepared to treat AI as a spiritual voice, and that the willingness climbs sharply among younger adults. Our congregations are forming a relationship with this technology whether we guide them or not. Which means the preacher who learns to use AI rightly is not only protecting their own study. They are modeling, for people already drifting toward AI as an authority, what a healthy relationship to the tool actually looks like.

The Conscientious Objectors

I do want to be fair to the resistance I did encounter, because it deserves respect, not dismissal.

At the end of my session, one participant raised her hand with a genuine concern: can we really embrace a technology that consumes so much water and carries an environmental cost? I told her, honestly, that I could talk for another half hour on that single question alone, and I pointed her to the fuller treatment I have already written on the subject.

I share that exchange because it points to something every minister and every educator needs to understand. There are people in our churches and our classrooms whom I have come to call “conscientious objectors” to AI. They are not Luddites. They are not afraid of technology. They hesitate because they believe, on principle, that it does harm, whether to the environment, to workers whose jobs are displaced, or to something they cannot quite name. That conviction is worthy of a pastor’s respect.

But here is what I have also learned: a great deal of the concern and hostility surrounding AI rests on misinformation, on misperceptions, or on metrics that are genuinely difficult to quantify and reconcile. That does not make the concerns illegitimate. It makes them a reason to ask better questions rather than to settle for easy verdicts. I have written about both the environmental impact and the question of AI and jobs elsewhere, and I would commend those pieces to you if these are the questions keeping you, or your people, on the sidelines:

• Asking Better Questions about the Environmental Impact of Technology - How to lead faithfully when technology debates become local

• AI Is Coming for Jobs. Now What? - What the data actually says — and why it matters for your congregation

After, Not Instead Of

So if the question is not whether but how and when, let me give you the most useful thing I shared in that room. It is a simple framework, and I offer it not as a rule but as a tool for your own discernment.

Think about your sermon preparation in three phases.

1. The first phase is listening. This is the front end of everything. Prayer. Reading the text slowly, in more than one translation if that is your practice. Sitting with the passage. Asking the Spirit to illuminate it. Letting your own pastoral context, the faces of your people, the burdens you have been carrying all week, come into contact with the text before anyone else’s framework does. Noticing what troubles you, what you do not understand, what you are drawn to, what you are resisting. This phase belongs to you and the Spirit. AI does not enter here.

2. The second phase is study. This is where you do the exegetical and interpretive work: original language study, commentary research, historical and cultural background, cross-references, theological reflection. This is where AI can be a genuinely useful research partner. Not the first voice in the room, but a capable assistant. Help me understand the major interpretive positions on this passage. What is the cultural background of this practice in first-century Judaism? What do Wesleyan commentators emphasize here? Those are legitimate, valuable uses.

3. The third phase is development. This is where the sermon takes shape: structure, illustrations, language, application. AI can help here, too. Not to generate the sermon for you, but to pressure-test your argument, to brainstorm illustration possibilities, to surface assumptions you are making about your congregation that may not hold.

The key principle is sequence. AI enters the process after you have listened, not instead of listening. After the Spirit has had first access to the preacher and the passage, not before. This is not about limiting AI. It is about protecting something irreplaceable.

And here is why that first phase cannot be skipped. There is a kind of preparation where you survey the passage, gather the information, identify the main point, and build a structurally sound message. It is efficient. It is organized. AI is extraordinarily good at it. And then there is another kind of preparation, where you sit with a passage long enough that it begins to trouble you, where the comfortable reading breaks down and you have to stay in the discomfort until something new emerges. The preacher who has been through the second kind of preparation carries something into the pulpit that the first kind cannot supply. And the congregation can feel the difference, even when they cannot name it.

AI cannot do the second kind. It cannot be troubled by a passage. It cannot wrestle, because wrestling requires something at stake, and AI has nothing at stake.

And this is not a knock on the technology. It is simply what the technology is. For all its fluency, generative AI cannot truly think, it has no empathy, it does not feel emotion, and it has no real understanding of the words it produces. It is making a remarkably sophisticated prediction about what word should come next, drawn from patterns in an enormous dataset. That is genuinely useful for the second kind of preparation. It is no substitute at all for the first.

The risk was never that AI would produce bad sermons. The risk is that it would produce sermons that were never wrestled with.

Not Just for Preachers, but for ALL Communicators

Now, I know not everyone reading this steps into a pulpit on Sunday morning. τέκτων has always been about the intersection of technology, innovation, and ministry, and that ministry takes a wide variety of forms.

So let me widen the frame without losing the point. The same sequence that protects the preacher protects the Christian educator preparing a lesson, the small group leader building a study, the worship leader planning a service, the communications director shaping a church’s message. You may never preach a sermon, but you are still doing the work of listening for what God is saying and then giving it form. And you face the same temptation the preacher faces: to let AI frame the encounter before you have done your own listening, and in doing so, to hand away the most valuable part of the work. The discipline is the same. Listen first. Bring AI in after.

Finding the Right Balance

How do you know if you have the balance right? In the workshop, I offered two short lists, and I will offer them to you here as a self-check.

There are signs you may not be using AI enough:

• Your research method has not changed in ten years. I still treasure my Strong’s Concordance, a gift from my parents for my seventeenth birthday, shortly after I was called to ministry and licensed to preach. But I rarely reach for the bound copy anymore, and that is not a betrayal of it. It’s deeply integrated into other digital tools I use now, like BibleHub’s Interlinear, Logos, and BlueLetterBible.

• You are not getting honest feedback. Most pastors are surrounded by people who love them and do not want to hurt their feelings. You won’t get honest feedback unless you explicitly ask for it, and push for it. An AI tool may initially be sycophantic, but if you tell it you want an honest critique, and you want tips for improvement or clarity, it will. It is not afraid to tell you the third illustration in a row assumed a two-parent household.

• You have decided AI is a threat without ever giving it a try. That is not discernment. That is merely avoidance, and it’s dishonest. The staunchest opponents of AI in ministry are those who have never actually used it or tried working with it. Once you do, it changes your approach from avoidance to discerning how to use it appropriately and effectively.

And there are signs you may be over-relying on AI:

• You reach for the AI before you spend meaningful time in prayer and silence. That is not a technology problem. It is a spiritual formation problem that the technology has merely exposed. AI has a place, but it’s definitely NOT the first place to turn.

• The sermon is starting to sound more like the AI than like you. Over time, the trend becomes hard to miss. Having it help you explore possibilities, consider nuances, draft outlines, come up with effective illustrations, etc., can be good, but if it’s just generating content for you and you’re not driving 90% of the thinking process, it will quickly show.

• You cannot reconstruct your own argument without consulting the draft it helped you generate. We see this in education circles, and it’s a clear sign that the AI did the work, not the user. When you are in the driver’s seat, you know what you are going to say and why, and you’re using the AI to hone and perfect your presentation and conclusions.

Hold those two lists side by side, and one question rises to the surface. It is the question I left that room with, and I leave it with you:

Is the way I am using AI, or not using it, making me a more faithful, Spirit-attentive, pastorally present preacher? Or less of one?

The Amplifier

Years ago, in some of the small rural Midwest churches I served, there was an old tradition of a receiving line. People would file past on their way out the door and offer the customary “Nice message, pastor.” It was kind and cordial, but as an introvert, I hated it; it always felt so inauthentic. And more often than I would like to admit, I wondered whether the person giving that compliment had actually heard a word I said, or whether they would be doing anything different on Monday as a result.

But this room was not like that.

When they invited me to lunch and then to a book table to sign copies and answer questions, I got to talk with the participants one by one.

I asked nearly everyone similar questions:

• What did you find most helpful?

• What was your biggest takeaway?

• What resonated most with you?

And every single person had an immediate, specific answer. There was no polite vagueness. They were interested, eager, open, and engaged. They wanted to know where the guardrails are, what works and what does not, and how to use this well. I found that deeply encouraging, and I think it bodes well for the future ministry of the Church.

Because in the end, that is what AI is. It is an amplifier. It amplifies what you bring to it: your voice, your perspective, your pastoral context, and the Spirit’s insight and urgency. It cannot supply what you do not bring. The sermon that changes a life this Sunday will come from a preacher who has prayed, wrestled, listened, and shown up. AI can absolutely help you prepare better, and more efficiently.

But it cannot preach for you. And quite frankly, it shouldn’t.

The Spirit still speaks. Our task is to make sure we are still listening, first.

If these are the questions you are wrestling with, I would welcome you as a subscriber to τέκτων (tektōn). It’s all about the intersection of technology, innovation, and ministry, with a special emphasis on responsible use of generative AI.

Resources referenced in this article:

Barna Group, in partnership with Gloo. “AI Is Becoming a Spiritual Authority, Even Among Practicing Christians.” Barna. May 19, 2026. Accessed June 7, 2026. https://www.barna.com/research/christians-trust-ai-flourishing-spiritual-authority/.

Chatterji, Aaron, Thomas Cunningham, David J. Deming, Zoe Hitzig, Christopher Ong, Carl Yan Shan, and Kevin Wadman. “How People Use ChatGPT.” NBER Working Paper 34255. National Bureau of Economic Research, September 2025. https://doi.org/10.3386/w34255.

OpenAI. “Signals Consumer Data.” OpenAI. May 11, 2026. Accessed June 8, 2026. https://openai.com/signals/data/.

Swisher, David J. “Redeemed for Those Not Yet a Part.” τέκτων (tektōn). May 11, 2026. https://tektonministry.substack.com/p/redeemed-for-those-not-yet-a-part

—. “Asking Better Questions about the Environmental Impact of Technology.” τέκτων (tektōn). April 27, 2026. https://tektonministry.substack.com/p/asking-better-questions-about-the

—. “AI Is Coming for Jobs. Now What?” τέκτων (tektōn). April 13, 2026. https://tektonministry.substack.com/p/ai-is-coming-for-jobs-now-what

—. Early Warning: How to Lead with Wisdom in Times of Technological Disruption. Invite Press, 2025.

Your phone rings, and it is an anonymous donor you have never met. “I’ve been impressed with the ministry you’ve been doing lately, but I also know you’re overworked and underpaid. I’d like to empower you to do more, so…if you’re willing, I would like to fund an assistant for you,” they say. “Someone skilled and reliable, who can absorb the logistics, the scheduling, the prep, the paperwork, etc. Someone who can handle all the tedious and time-consuming things that drain your week, so you can pour yourself into the work you were actually called to do.”

Most of us would say “yes” before they even finished the sentence.

But then comes the second question, and this is the one that actually matters: “So tell me…what, exactly, would you hand off (what would you delegate to that assistant)?”

Take a moment with that. Not in a rhetorical or hypothetical sense, but for real:

• If a capable assistant walked into your office tomorrow morning and asked you to name everything you would gladly take off your own plate, what would make the list

• If a trained intern showed up at your church and offered to offload a good part of your weekly burden so you can focus on more effective ministry, what would you have them do?

When I pose this question to ministry leaders, the answers come quickly. People inevitably name the tedious, the monotonous, and the repetitive. That’s the work that saps their time and energy without requiring their professional training, or their pastoral relationship with the people they serve. Nobody hesitates to hand off the parts of the job that drain them, the stuff that never feels like the reason they answered a call to ministry.

With generative AI now widely and affordably available, that office assistant is essentially already on offer. And the question of what to hand off turns out to be the most useful question you can ask about AI in ministry.

Subscribe now

This Is a Vocation Question, NOT a Productivity One

It would be easy to read all of this as a time-management or efficiency pitch. Do more, faster. Clear the inbox in half the time. And yes, AI can do that. However, that framing misses the point entirely.

The reality is that the goal is not speed for its own sake, and it is certainly not eliminating your workload so you can simply do more of everything. The goal is to protect your capacity for the work that only you can do.

That single sentence is the governing principle behind all of this, and it is worth reading again, slowly and very deliberately:

Faster sermon formatting is not the win. The win is the extra hours of prayer, study, and Holy Spirit presence that the faster formatting makes possible.

Think of it as a kind of stewardship. Your time and attention are finite, and they are holy resources. Every hour you spend wrestling a bulletin into shape is an hour not spent with a grieving family, or sitting with a difficult text, or simply being available when someone needs a pastor. When AI carries the logistical weight, it is not making you more productive in the abstract. It is giving your calling more room to breathe.

So the question is not really “How do I get more done?” The better question is “What is the work that bears my hand, and how do I clear space to do it well?”

What You Would Gladly Hand Off

Whatever you would have happily given to that donor-funded assistant is a strong candidate for AI. In ministry, that list is longer than most leaders realize:

• Drafting routine emails and replies.

• Building first drafts of bulletins, slides, newsletters, and orders of service.

• Coordinating schedules and untangling calendar conflicts.

• Transcribing a meeting and turning it into a clean summary with action items.

• Generating a starting outline, a set of title options, or a handful of illustrations to react to.

• Gathering background and cross-references as a research starting point.

• Turning a finished sermon into small-group questions, a study guide, a Kids’ church activity, or a newsletter recap.

• Proofreading and tightening your prose.

• Producing accessible materials, alt text, and translations.

• Spotting trends in attendance or giving data that would take you hours to surface by hand.

Notice the common thread? None of these tasks requires your ordination or calling! None of them depends on your relationship with the person in front of you. They are real work, and they matter, but they are not the work that only you can do. They are exactly the kind of thing a skilled assistant would handle, which is precisely why they are ripe to delegate.

What You Should Protect

Now the harder column, and the more important one.

The stuff you said really needs to be you, and only you? That is the work you guard. That is the work you insist on doing yourself, even when an algorithm offers to take it off your plate.

Here is where this lives:

• The actual proclamation of the Word, filtered through your own prayer and discernment.

• The hospital visit.

• Sitting with the grieving.

• The hard conversation that needs your presence and not a polished message.

• Spiritual direction and counsel.

• Genuine relationship and the slow work of shepherding.

• Your own spiritual walk with God, which no tool can outsource.

• And the final theological judgment about what is taught under your care, which belongs to you and should never be outsourced to a model.

Where this line falls will understandably vary from one religious tradition to another. A sacramental tradition reserves things that a low-church one would not, and what one community entrusts only to the ordained, another shares more broadly. So, I am not going to draw that line for your context; you and your community know your specific contexts and expectations better than I do. But every tradition has a core that bears your calling, your relationship, and your discernment, and that core is not for hire.

It is worth remembering, too, that this is not only a theological boundary. It is a practical one. I have written before about how AI consistently underperforms in the relational dimensions of ministry and learning, the very places where a present, attentive human notices what is unsaid and chooses to stay in the room. The line you protect is also, conveniently, the line where the technology is weakest. Guarding it is not nostalgia. It is simply good, practical ministerial sense.

Delegation, Not Abdication

Let me be clear about something else, because it is the place where this whole idea most often goes wrong: letting a generative AI tool help with the logistics and tedious tasks is NOT walking away. It is NOT abdicating responsibility; it is delegating it.

An assistant, even a brilliant one, still works under your oversight. You would never hire someone as an office manager, executive assistant, or ministry intern and then stop reading what went out under your name. No, as you’re getting to know their capabilities and strengths, you’re reading and scrutinizing everything. The same holds here, and it holds with extra force, because these tools carry a kind of baked-in formation of their own, a learned set of leanings shaped by how they were trained. I wrote more about this in another recent Substack article I posted here on τέκτων (tektōn). That is all the more reason to keep your hand firmly on the wheel.

This is exactly the tension I explored in Early Warning. In Chapter 2, I introduced Pastor Rob, who uses generative AI much like a colleague or administrative assistant, asking it for outline help, illustrations, research leads, and follow-up study materials, while never letting it write the sermon for him. His work improves, and he saves hours. But when he mentions it to his leadership team, some of them push back, and their concerns are good ones.

They ask how much of the sermon is truly his if the AI is helping shape the content. They wonder whether leaning on AI dilutes the spiritual discernment that has always been part of pastoral preparation. They ask how the congregation would feel if they knew. And they raise the sharpest question of all: even if Rob’s own use is perfectly appropriate, is there a line that someone following his example might cross?

Here is why those questions matter. The instinct behind every one of them is correct. They are oversight questions, and a leader who never asks them is the one in danger.

Rob’s response is a great model, though: he does not abandon the tool, and he does not pretend the concerns are silly. Instead, he decides to use AI as a starting point while committing that every sermon will still be filtered through personal prayer and reflection, and he proposes being open with his congregation about the role of technology in his work rather than quietly substituting it for his own labor.

That is the whole posture I’m advocating here, in miniature:

This is not merely a tidy resolution on the page. In fact, I do not have to look far for a living example. My own pastor uses AI in drafting his sermons, and I am convinced it has made him a better preacher, not a lazier one (he would readily concur, too). The way he describes it, AI carries the logistical burden and the drudgery so that he can spend more time percolating with the text, giving the Holy Spirit room to work on him and through him.

Read that again, because it is the entire argument in a single sentence from a working pastor. The tool did not replace the all-important spiritual work. Rather, it cleared space for more of it.

Making Your List

So what does this mean for you, practically?

Do the donor exercise for real. Over the next few days (or ideally a full week), keep a running list of every task that crosses your desk, and sort each one into two columns. In the first column, put everything you would gladly hand to a capable assistant. The stuff that keeps you busy, but doesn’t require any special ministry calling, training, or theological skills. In the second, put the work that needs to be you, and only you.

That second list is sacred. Protect it.

That first list is your AI roadmap. Start with one or two items, learn the tool, build a little trust, and keep your oversight engaged the whole way. You are not trying to automate your ministry. You are trying to give it back the time it has been quietly losing to busywork.

The Work That Bears Your Hand

There is a reason this publication is named for the word τέκτων, the artisan and craftsman that Jesus was. A craftsman delegates plenty. He does not mill his own lumber or forge his own nails. But there is work in every craft that bears the maker’s own hand, the part you would never subcontract because it is the very thing that makes the work yours.

Ministry has that work too. Let the tools carry what they can carry. Then spend what they give back on the work that no algorithm will ever do for you.

Resources referenced in this article:

Swisher, David J. Early Warning: How to Lead with Wisdom in Times of Technological Disruption. Plano, TX: Invite Press, 2025. See especially chap. 2.

Swisher, David J. “The Bot Nobody Asked For: What Khanmigo’s Demise Tells Ministry Leaders About AI & the Limits of Simulated Presence. τέκτων (tektōn). May 4, 2026. https://tektonministry.substack.com/p/the-bot-nobody-asked-for

Swisher, David J. “Can Computers Form Habits? A Peek Inside the Recent “Goblin” Problem in ChatGPT. τέκτων (tektōn). May 25, 2026. https://tektonministry.substack.com/p/can-computers-form-habits

I recently returned from several days of meetings and conversations with ministry leaders from across the Wesleyan world at our quadrennial General Conference, connecting with pastors, educators, district superintendents, and denominational leaders. The conversations ranged widely, as they always do at gatherings like this.

But whenever the topic turned to technology and artificial intelligence (and it went there often), I kept hearing the same three things:

• The first was a sense of anxiety. It wasn’t an outright fear or panic, but a quiet, persistent unease. It’s a sense that something significant is happening, that the pace of change is accelerating, and that the clarity needed to lead well simply isn’t there yet. When clarity is absent, hesitation fills the gap.

• The second was a cautious worry about identity and calling. Most leaders I talked with were genuinely interested in what generative AI can do, but equally concerned about relying on it too heavily. If we let AI help draft our content, answer our emails, plan our services, or assist with sermons, are we slowly handing off something essential? The question underneath that worry is a real one: where is the line between using a tool and being used by it?

• The third was this: everybody is talking about AI, but only a limited number of people (especially in ministry contexts) are actually providing clear guidance. The conversation is everywhere, but the wisdom is not, and I heard this sentiment a LOT.

I started τέκτων (tektōn) three months ago because I believed those three things were true, and that we needed a place in the church to talk candidly about the intersection of technology, innovation, and ministry, with a special emphasis on responsible use of generative AI. Those conversations confirmed it.

Subscribe now

What Three Months of Writing on Substack Has Taught Me

I’ve published thirteen articles now, and looking back across them, several have really stood out in the conversations they generated and the questions they answered:

1. A Better Way to Think About Technology in Ministry and Centered-Set Thinking and the Ministry Technology Question were designed as a pair, and together they have resonated more than anything else I’ve written. The core argument was simple but powerful: most ministry-oriented conversations about technology default to binary thinking without realizing it: (1) AI is either going to save the church or destroy it. (2) That new platform is either a gift from God or a threat to be resisted. I applied Paul Hiebert’s centered-set framework, and it breaks that false choice open and replaces it with a better question: does our use of this technology move us toward Christ, or away from him? That question doesn’t make the hard decisions disappear. But it gives you a much better map for navigating them. If you haven’t read those two pieces, I’d recommend you start there.

2. The Bot Nobody Asked For took on the question of appropriate use directly, and it was shared and commented on quite a bit. I analyzed the demise of Khanmigo, the AI tutoring tool that was supposed to change education, as a lens for what happens when we ask technology to do something it fundamentally wasn’t designed to do. The parallel for ministry is direct and sobering. I spent a significant portion of that article thinking through who is most likely to seek out AI pastoral care in the first place, and why. The answer reframes the whole conversation. It’s not an abstract ethics question. It’s a question about real people in real crisis, and what they actually need from the church.

3. Redeemed for Those Not Yet a Part went back to 1793, to a Cambridge pastor who faced enormous resistance when he introduced a new technology into his congregation’s worship. I’ve stood in that church, and seen the barrel organ still sitting there, and I loved worshipping with that still-alive congregation. But the resistance that Charles Simeon experienced was fierce, and the anxiety his parishioners felt was real. The parallels to our current moment are striking. Leaders who engage wisely with technology, rather than simply digging in or surrendering, are the ones history remembers.

4. Preaching with Two Owls? tackled the identity and calling question head-on for preachers, drawing on doctoral-level work in homiletics that my friend and mentor Leonard Sweet and I have been developing together. Can a machine preach? The question is no longer hypothetical. But the answer is more nuanced than a simple yes or no, and it has everything to do with what we believe preaching actually is. I highlighted the rationale for the upcoming cohort that Len Sweet and I are developing to explore semiotics and AI together (and yes, there’s still room for a few motivated students to join us).

5. AI Is Coming for Jobs. Now What? took on the broader anxiety that’s driving so many of these conversations. The disruption is real, and I take the research seriously rather than either dismissing the concern or amplifying the panic. The growing worries over AI-related job losses and disruption is a significantly growing pastoral concern we need to be paying attention to and able to answer. That article doesn’t offer false comfort. It offers the kind of honest, grounded engagement I hope characterizes everything I write here.

A Request from Me to You

You are reading τέκτων (tektōn) because somewhere along the way, one of these conversations found you, most likely through a post on Facebook or LinkedIn. Regardless of how you found it, I’m glad you’re here!

The open rates and readership have been genuinely encouraging, and the feedback I’ve received has been affirming. In fact, the open rates have consistently exceeded industry benchmarks by a significant margin. But with only 25 subscribers, these conversations are reaching a much smaller audience than they should be. That’s where you come in.

If any of these articles have been useful to you, I have a simple ask:

Pick one – whatever article resonated with you most, one you found yourself thinking about later, the one you’d want a colleague to read. Then Restack it on Substack, or share the link directly with one specific person you know who is wrestling with these same issues and questions.

No need to broadcast it to dozens. Just one person (or even a few) - you probably already have one or more people in mind.

After all, every one of these articles is still timely.

The questions I kept hearing reminded me that ministry leaders are hungry for this. They’re not looking for hype or alarm. They’re looking for wisdom, a framework, and a way forward that honors their calling and takes the technology seriously. That’s exactly what I’m here to provide, week after week. τέκτων (tektōn) is free and posts every Monday morning.

If you know someone who needs it, now would be a good time to make the introduction.

Something strange began appearing inside ChatGPT.

Starting with GPT-5.1, OpenAI’s models began inserting an odd kind of language into their responses. The word “goblin” started showing up in places it had no business being. Then “gremlin.” Then trolls, ogres, raccoons, and pigeons. References to these creatures began appearing inexplicably across millions of conversations. By GPT-5.4, goblin mentions had increased by nearly 3,900% within ChatGPT’s “Nerdy” personality mode.

It took months before anyone at OpenAI could explain why.

When they finally traced the cause, they patched it the only way they could in time for their next model release: they buried a line in the system prompt.

“Never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user’s query.”

Here it is in context:

Notice what that is, and what it isn’t. It isn’t a fix. It’s an instruction to suppress a behavior that had already been baked in too deeply to retrain out in time. A multi-billion-dollar AI lab, telling its model to stop summoning goblins...because they couldn’t fully undo whatever had caused it.

The goblins, it turns out, are an excellent teacher. Because what OpenAI documented, without quite using this language, is that their AI had formed a habit.

How It Happened

Let’s explore briefly, and non-technically, how these models are trained, because it matters for what comes next.

OpenAI, like most major AI developers, uses a process that involves two key stages:

• In the first stage (called “pre-training”), the model is trained on vast amounts of text and learns to predict what words and phrases naturally come next in context.

• In the second stage, human raters score the model’s outputs, and the model is adjusted to produce more of what earns high scores and less of what earns low scores. This scoring process is called reinforcement learning.

OpenAI had been developing a personality customization feature for ChatGPT, and one of those personalities was called “Nerdy.” Its instructions told the model to be “unapologetically nerdy, playful, and wise,” to “undercut pretension through playful use of language,” and to acknowledge the strangeness of the world. Somewhere in that training process, creature-language, goblins, gremlins, the whole menagerie, kept appearing in the outputs that scored highest.

The model learned that in playful, nerdy-toned conversation, creature metaphors got rewarded. So it produced more of them. Those rewarded outputs were then fed back into the next round of training, which produced more goblins…and those scored well again. And so on.

OpenAI’s own postmortem described the pattern plainly: once a style tic is rewarded, later training can spread or reinforce it, especially when those outputs are reused in subsequent training data.

The behavior didn’t stay contained to the Nerdy personality. It spread. Eventually, the model was producing goblins in conversations where no Nerdy context was present at all. The behavior had generalized, and now it was self-reinforcing.

Sound familiar?

Subscribe now

So, actually… Yes. Computers Can Form Habits.

If you have read Charles Duhigg’s The Power of Habit, the pattern described above will be immediately recognizable.

Duhigg’s research identified a three-part loop at the core of every habit: a cue, a routine, and a reward. The cue triggers the behavior. The routine is the behavior itself. The reward is what reinforces it and makes the loop run again. What makes this framework so powerful is that it explains not just how habits form, but how they persist, how they generalize, and how they can become nearly automatic over time.

The goblin habit formed the same way.

The cue wasn’t the Nerdy system prompt itself; rather, it was the broader training environment of playful, nerdy-toned conversation. That was the context in which creature-language reliably scored well. The routine was the goblin and gremlin metaphors. The reward was the high training score. And once that loop ran long enough, the behavior stopped requiring its original context and began showing up across other personality types entirely, without the original context.

I can illustrate this from personal experience. For years, I had an all-too-predictable habit of falling asleep on the couch in the late evening, regardless of how engaging what I was watching was. I’m sure the first few times I was truly tired. But after the habit loop formed, I’d fall asleep on the couch, even if I wasn’t tired…because it had become a habit with a predictable routine and reward. After reading Duhigg’s book, I tried a small experiment: I moved where I sat from one end of the couch to the other. Six feet. Nothing else changed. Same couch, same time of night, same shows. But the habit broke completely.

Why? Because the cue that had been triggering the sleep routine wasn’t the time of night, or the show, or even the couch itself. It was the specific spot on the couch. Once the cue was removed, the loop would no longer activate.

OpenAI’s “fix” worked on the same principle, except in reverse. Instead of removing the cue, they added an instruction to override the routine. The habit was still there, baked into the model’s training. They just told it to ignore the impulse. That is not a solution. That is management.

James Clear takes this insight a step further in Atomic Habits. The most durable habits, he argues, aren’t just behaviors we practice repeatedly; over time, they become part of how we see ourselves. We stop performing a behavior and start being the kind of person who naturally does it. The habit becomes part of our identity; it becomes ingrained into who we are.

What OpenAI’s postmortem documents is precisely that kind of drift. The model didn’t just pick up a verbal tic. It developed a default lean, a habitual tendency, a kind of character, shaped by what it had been consistently rewarded for producing. Nobody designed a creature-language model. The model became one through accumulated reinforcement.

We become what we practice. And apparently, so do AI models.

But Who Decided What Gets Rewarded?

The goblin problem is funny, but the broader pattern it represents is not. Here is where the goblin story stops being funny.

Who decided that playful and nerdy were qualities worth rewarding at high scores? Who defined what a “quirky” or “good” response looks like? Who were the human raters whose preferences shaped the model’s emerging personality?

These are not rhetorical questions. They have answers, and those answers matter enormously for ministry leaders who are inviting these tools into their work.

Amy Webb, in her essential book The Big Nine, traces the history of AI development and the people who built it (see Ch. 2 - The Insular World of AI’s Tribes). The group of computer scientists who gathered at Dartmouth in 1956 and coined the term “artificial intelligence” were all white men, all university-educated, all in their forties at the time. The universities that became the centers of AI research and received the most funding from corporations and government were, broadly, secular and progressive institutions: Stanford, Carnegie Mellon, Harvard, UC Berkeley, Cornell, MIT. And because these were hard science fields in the 1960s and ‘70s, the voices most underrepresented in both the research and the early training datasets were women, minorities, and conservatives.

There are exceptions, of course, and the picture is not monolithic. There have been many newer training sets of data, some of which I have used. But across the broad spectrum of AI development as a whole, that is the heritage from which AI bias emerges. Many of my educator colleagues have observed and documented this firsthand: if you ask AI to illustrate a doctor or scientist, it will most likely generate a white male, even though there are hundreds of female and ethnic doctors and scientists. If you ask it to illustrate a nurse, there’s a high probability it will depict a white female, even though male nurses and nurses from most ethnicities exist in abundance. Why? The absence of such examples in the training data, plus reinforcement learning.

We can, and should, specify what we want to see through our prompting, and be intentional about how we represent diverse people, views, and roles. But part of our responsibility in managing AI well, especially as ministry leaders, is calling technology providers to do better, to improve their models so that the full spectrum of humanity is represented, and so that conservative and religious viewpoints are well-represented too, not just an afterthought.

What this means practically is that the “personality” baked into these tools reflects a particular cultural and ideological formation. It’s NOT malicious, and it’s not a conspiracy. It is simply formation, the same kind of formation that shapes any person or institution over time. You can work around it, and you often should, but you have to know it’s there first.

The Pastoral Cost of Not Knowing

OpenAI published a thorough postmortem once they understood the goblin problem. That transparency is worth acknowledging. But the timeline is equally worth noting: the behavior began with GPT-5.1, intensified across multiple model generations, and the public explanation only appeared after the system prompt fix had leaked on GitHub and spread across the internet. That is NOT proactive transparency. That is disclosure after the story had already broken.

In the months in between, millions of users were interacting with a model behaving in ways its own creators did not fully understand. That is anything but ethical. If you want to see an opposite ethical approach, look at how Anthropic (maker of Claude) regularly and intentionally creates hypothetical scenarios to see what could go wrong, and isn’t afraid to share those results, even when the outcomes are bad (and when one of their planned models became so good it found exploits that computer programmers hadn’t identified in decades, they pulled it from production). That’s responsible AI in action.

Consider a parallel case that may already be familiar. In February 2023, the day after the mass shooting at Michigan State University, Vanderbilt University’s Peabody College Office of Equity, Diversity, and Inclusion sent a campus-wide condolence email to students. At the bottom of the email, in smaller print, was a parenthetical: “Paraphrase from OpenAI’s ChatGPT AI language model, personal communication, February 15, 2023.”

They did disclose it. Technically. But only in a footnote.

The student newspaper caught it. The backlash was immediate and significant. The dean of the college issued a public apology, acknowledging that the email “so missed the crucial need for personal connection and empathy during a time of tragedy.” Two associate deans temporarily stepped back from their positions.

Notice: The problem was NOT that AI was used. The problem was that it was used in a moment that demanded genuine human presence, and the only sign anyone had considered that question at all was a citation buried at the bottom of a condolence letter like a legal disclaimer.

Now imagine that email is a pastoral letter from a district leader to a congregation after a tragedy. Or a message from a campus minister to students after a loss in the community. The stakes would be nearly identical: the trust you would lose is not recoverable with a follow-up explanation.

Ministry leaders bear the accountability for how they deploy these tools. The AI company’s postmortem, whenever it arrives, does not transfer that responsibility.

So What Do You Do With This?

None of this is an argument against using generative AI in ministry. It is an argument for understanding what you are actually working with.

A few questions worth sitting with before you integrate any AI tool into your ministry workflow:

• Do you know what the developers themselves say about how this tool was trained and what it’s designed to optimize for? Most major AI developers, including OpenAI and Anthropic, publish documentation called model cards or system cards that describe training approaches and known limitations. They’re not light reading, but they contain exactly the kind of information this article has been exploring. The goblin postmortem itself is a prime example: it was publicly available for anyone who went looking. Start there.

• Can you recognize when this tool gets things confidently wrong? Generative AI doesn’t know when it’s guessing. It predicts what words come next in context, and sometimes that prediction is flagrantly incorrect, delivered with complete confidence. Here’s one of my favorite examples of this:

Clearly, verification is not optional!

• Are you using it in a context that requires genuine human presence? AI handles many tasks beautifully: drafting, research, brainstorming, administrative efficiency, sermon preparation support. There are also moments where substituting a prediction engine for a human voice is a pastoral failure, no matter how polished the output looks.*

• And are you being transparent (genuinely transparent, not footnote-transparent) about your use of AI with the people you serve? If AI is being used in a significant contributory way, that should probably be disclosed.

Wise Workmanship

The tektōn, the craftsman, the builder, knows what materials they are working with. They understand the grain of the wood, the composition of the stone, where the material is strong and where it is prone to failure. That knowledge doesn’t make them afraid of the material. It makes them competent to use it well.

The goblins are a parable about what happens when we don’t ask those questions. A capability was introduced, a training process ran, incentives accumulated, and a behavior emerged that nobody intended, but it went unaddressed, worsening exponentially, for a full six-plus months. The behavior had character. It had habits. It had a formation history.

The tools you are considering for your ministry have a formation history, too. The wise workman learns it and constantly monitors what it’s forming.

* In my training presentations for university faculty and ministry leaders, I discuss examples of when AI should never be used; hopefully a future article can explore that.

Resources referenced in this article:

Clear, James. Atomic Habits: An Easy and Proven Way to Build Good Habits and Break Bad Ones. New York: Avery, 2018. https://amzn.to/4dikhNd

Duhigg, Charles. The Power of Habit: Why We Do What We Do in Life and Business. New York: Random House, 2012. https://amzn.to/4wDjCxs

Korn, Jennifer. “Vanderbilt University Apologizes After Using ChatGPT to Write Mass Shooting Email.” CNN Business. February 22, 2023. https://www.cnn.com/2023/02/22/tech/vanderbilt-chatgpt-shooting-email

OpenAI. “Where the Goblins Came From.” OpenAI. April 29, 2026. https://openai.com/index/where-the-goblins-came-from

Webb, Amy. The Big Nine: How the Tech Titans and Their Thinking Machines Could Warp Humanity. New York: PublicAffairs, 2019. https://amzn.to/4nyVWWQ

#AffiliateLinks - for more about my use of affiliate links, see the “About” page.

You’ve Landed in the Right Place

If you’re a pastor, ministry leader, or anyone who cares about using technology wisely in Christian ministry, this publication was written for you. The noise around artificial intelligence and digital tools is genuinely overwhelming right now. The hype is loud, and the fear is real. Finding practical, grounded, biblically-informed wisdom in the middle of all of it is much harder than it should be.

That’s the gap τέκτων was built to fill.

Why “τέκτων”?

The name comes from the Greek word used in the New Testament to describe Jesus’ occupation. Most translations render it as “carpenter,” but that’s an oversimplification. In the arid Mediterranean world where Jesus lived and worked, stone was the norm, and wood was scarce. So, linguistically and archaeologically, Jesus was almost certainly a stonemason. More importantly, he was a craftsman…a builder who worked with the best available materials of his day.

In other words...he was a technologist.

If Jesus is our model as τέκτων, then following his example means engaging with the tools of our era with skill, purpose, and wisdom. Not fear. Not reckless adoption. Wise, purposeful workmanship.

For a deeper look at this tektōn concept, see Chapter 2 of Early Warning (pp. 41-45).

Who I Am

I’m David Swisher, a ministry technologist and ordained Wesleyan minister with 39 years of bivocational ministry experience. I serve in a consulting role as Director of Online Services at Kingswood University (Sussex, New Brunswick), and I provide training and consulting on AI and technology integration for ministry leaders and higher education faculty. My book Early Warning: How to Lead with Wisdom in Times of Technological Disruption (Invite Press, 2025) covers many of the themes you’ll find developed further here each week, and in this Substack, I dive in deeper on them.

What You’ll Find Here

τέκτων posts every Monday. Articles fall into five categories: (1) deeper dives on themes from Early Warning; (2) practical use-case examples you can actually put to work; (3) review and discussion of tools, practices, and guardrails; (4) analysis of emerging technology trends with a ministry lens; and (5) reviews of books and resources on AI and ministry. Every article aims at the same target: equipping you to use technology faithfully, wisely, and without getting lost in the noise.

Where to Begin

With a full archive of articles already available, here are a few natural starting points depending on what brought you here:

• If you want a framework for thinking about AI in ministry, start with A Better Way to Think About Technology in Ministry and Centered-Set Thinking and the Ministry Technology Question. These two articles lay out a practical theological lens for evaluating any technology you encounter.

• If you’re a preacher or someone who thinks deeply about ministry history and innovation, Redeemed for Those Not Yet a Part traces how one pastor’s controversial embrace of new technology reshaped a congregation, and what that story still has to say to us today.

• If you’re wrestling with the real costs and risks of AI, AI Is Coming for Jobs. Now What? and Asking Better Questions about the Environmental Impact of Technology take an honest look at what responsible adoption actually requires, and explores the pastoral care impact of technology consumption and AI deployment.

• If you want a practical tool review, The New Front Porch: How the Pandemic Rewired Church Attendance or The Tool That Showed Me How to Pay Attention show what thoughtful, responsible technology use looks like in real ministry contexts.

Free or Paid: Both Are Welcome

Everything in the weekly Monday articles is free. Paid subscribers receive additional bonus content, early announcements, and advance dialogue on projects in development. Either way, I’m glad you’re here.

If this is the kind of thinking you want showing up in your inbox every Monday, subscribe below. It’s free to start, and I’d love to have you reading along.

Subscribe now

It is no longer a question of whether AI will affect your church. It already has. Your staff is experimenting with it. Your volunteers are using it. Your congregants have opinions about it. What most ministry teams are still missing is a framework, and Jason Moore has written the best one available.

AI and the Church: A Clear Guide for the Curious and Courageous (Invite Press, 2024) is 298 pages of accessible, theologically grounded, and practically useful guidance on artificial intelligence written specifically for church leaders. Whether your team is already experimenting with AI tools or still working out whether AI belongs in ministry at all, this book meets you where you are and takes you somewhere better.\

Why I’m the Right Person to Review This Book

Jason Moore is not simply an author whose work I am assessing from a distance. He is a professional colleague and a friend. We have been talking about technology and ministry innovation for well over a decade. I served on his Book Launch Team for both this book and his earlier Both/And book on hybrid worship design. He has invited me to be a guest speaker with his AI training cohort. I wrote the endorsement that appears in the published edition. I am mentioned by name in the Acknowledgements.

I have been following the development of artificial intelligence for more than 30 years, and since ChatGPT’s arrival I have tracked and reviewed every significant book released on AI in ministry and church contexts. I now own more than 75 books on AI and machine learning, and I maintain curated annotated bibliographies on Raindrop.io for leaders who want a vetted starting point.

And the connection goes deeper than that. The book’s origin story runs directly through us. Jason describes it in the Acknowledgements: When my colleague Tiffany Snyder, director of faculty enrichment at Indiana Wesleyan University, was coordinating a Faculty professional development workshop on Generative AI, I recommended that we invite Jason to keynote it. That conversation, Jason writes, was the moment his enthusiasm for AI moved from a hobby to a professional calling. The speaking and training work we helped set in motion led directly to this book.

I am telling you all of this not to apologize for my perspective, but to establish it. I am likely the most qualified reviewer this book could have. I have read widely in this field, tracked its development carefully, and watched this particular book come to life from the inside. My recommendation carries the weight of that investment. When I tell you this is the best guide available for ministry leaders navigating AI, I am speaking as someone who has read the competition.

Subscribe now

Who Is Jason Moore?

Jason Moore is a graduate of The Modern College of Design and the author of 14 books. He has spent more than two decades designing resources, training programs, and consulting engagements for churches of all sizes, styles, and denominations. His background is visual communication and church creative arts, which gives him a distinctive angle on AI: he approaches it not as a technologist but as a practitioner who cares deeply about creativity, communication, and mission.

When the ChatGPT wave hit in the fall of 2022, Jason did what he does best. He dug in, learned everything he could, and started equipping churches. He has since spoken at multiple conferences on AI and the church, and he is actively training congregations and denominations across the country. Jason does not just write about AI. He shows up in rooms with ministry leaders and helps them use it wisely.

What This Book Actually Does

AI and the Church is organized into four parts that build logically on one another, and the structure reflects genuine pedagogical thoughtfulness.

• Part I demystifies AI: what it is, where it came from, and how it already shows up in tools your church is probably using. Moore explains complex concepts without technical jargon, with the patience of someone who has sat with too many ministry leaders who got lost in the weeds.

• Part II builds the theological and ethical foundation, and this is where the book earns its place on every ministry leader’s shelf. Moore addresses the real concerns head-on. Is AI the antichrist? Will it replace pastoral ministry? What are the ethical limits? He does not dismiss these questions. He takes them seriously, wrestles with them biblically, and offers a grounded framework for discernment. His argument from the Creation Mandate in Genesis 1, tracing the connection between God’s creativity and the human creativity that drives technological innovation, is worth the price of the book all by itself.

• Part III walks through the practical AI toolbox: chatbots, image generation, AI video, and more. These chapters are current and aimed squarely at practitioners rather than programmers.

• Part IV brings it all home with ten specific ways to use AI in the church, a look at where AI is headed, and a full sample AI policy your church can adapt and use immediately. That last chapter alone will save leadership teams hours of starting-from-scratch work. It is also where Moore distills his entire philosophy into three sentences that have stayed with me:

“AI should be a do it with you, not a do it for you. Working with AI should be a conversation, not a query or command. You have a soul. AI doesn’t. If you don’t bring your soul to the conversation, there is no soul in the conversation.”

— Jason Moore, AI and the Church (Ch. 12 - pg. 246)

Accessible, theologically grounded, and immediately applicable. Those three sentences function as a guiding principle for everything else the book teaches, and they reflect a pastoral wisdom that goes well beyond typical technology guidance.

My only disappointment is that AI and the Church is not currently available in audio format. For those of us who prefer audiobooks, that is a real gap. But don’t let that stop you from picking up the print or Kindle version. The content is too valuable to wait.

Why This Matters for Ministry Leaders

Here is the reality most ministry teams are navigating right now. AI is not waiting for the church to develop a theology of it. It is already in the staff meeting, the sermon prep workflow, the social media calendar, and the bulletin. The question is not whether to engage it. The question is whether you will engage it thoughtfully or reactively.

Most of the resources available on AI fall into one of two camps: enthusiastic tech promotion that glosses over the risks, or fearful resistance that offers nothing constructive. Moore occupies a much more useful third position. He is an advocate who takes the concerns seriously. He is a practitioner who stays grounded in Scripture and theology. He is a trainer who knows what actually works in a ministry context.

That balance is exactly what the church needs right now. The skeptic gets a fair hearing and a biblical case for engagement. The enthusiast gets the ethical guardrails they probably need. The pastor who just wants a practical starting point gets a clear path forward.

Moore’s three core principles deserve to become the starting point for your church’s AI conversation. If you lead a staff team, read those three sentences aloud at your next meeting and ask: “Are we doing this?” The discussion that follows will be worth far more than the time it takes.

And if your church has not yet developed an AI policy, Chapter 12 includes a full sample policy you can adapt to your context. It is one of the most practically useful contributions in the book, and it reflects the same balanced, biblically grounded approach that runs throughout.

AI and the Church is accessible enough for those with no technical background, yet substantive enough for those already experimenting with AI tools. It is written for practitioners, not programmers. Pastors, ministry staff, denominational leaders, active volunteers, and anyone asking “Is AI okay for the church to use?” will find a clear, trustworthy, and genuinely helpful guide here.

What You Can Do Next

Get a copy of AI and the Church. Better yet, get copies for your entire leadership team and your active media ministry volunteers. This is not a book to read in isolation; it is a framework to work through together, and Moore includes reflection questions at the end of each chapter that are genuinely useful for group conversation. Invite Press offers discounts on volume orders purchased directly through the publisher, so equipping your whole team does not have to break the budget.

If you are building out a broader reading list on technology and ministry, I maintain an annotated bibliography of vetted resources on Raindrop.io that I update regularly. You will find AI and the Church listed there. [AI and Ministry - Resources]

As I explain in Early Warning, the church has navigated technological disruption before— the printing press, the microphone, the internet. Each wave brought real risks and real opportunities, and the communities that engaged thoughtfully were better positioned to serve their neighbors well. AI is that kind of moment. Jason Moore has given us an excellent guide for navigating it.

Don’t wait for the AI conversation to come to you. Get ahead of it.

Disclosures: I served on Jason Moore’s Book Launch Team for AI and the Church and contributed an endorsement that appears in it. The Amazon links in this article are affiliate links (see the About page for more info).

Book Details and Links

Moore, Jason. AI and the Church: A Clear Guide for the Curious and Courageous. Plano, TX: Invite Resources, 2024. 267 pages. Available in paperback and Kindle (no audiobook edition).

My Curated List

Want more insights on useful books on AI for church and ministry contexts?

I maintain a curated list on Raindrop.io of all of the available books on AI for pastors and ministry leaders. It is available at https://WildcatTech.raindrop.page/ai-for-church-and-ministry-leaders-65707585. To the best of my knowledge, nobody else has anything comparable.

Resources referenced in this article:

Moore, Jason. AI and the Church: A Clear Guide for the Curious and Courageous. Plano, TX: Invite Press, October 2024. 298 pages. Paperback and Kindle.

• Publisher: https://www.inviteministries.org/store/ai-and-the-church-paperback.

• Amazon: https://amzn.to/49BTCb6.

Swisher, David J., comp. "AI and Ministry” [An Annotated Bibliography]. Raindrop.io. Accessed May 14, 2026. https://app.raindrop.io/my/65707585.

In the summer of 2013, I had the opportunity to visit Holy Trinity Church in Cambridge, England. My doctoral cohort was studying at Wesley Hall there and learning from the “pioneers” (innovative church planters in post-Christian Europe), and on Sunday, I suggested - and our resident Anglican heartily endorsed - worshipping there. I knew there was some history there, but was delighted by what I encountered.

I had read about Charles Simeon. But reading about a place and standing inside it are two different things, and what I encountered inside that five-hundred-year-old sanctuary was invigorating. The congregation wasn’t just a bunch of elderly saints still holding onto tradition; instead, it was filled with people of all ages, and especially young people. The worship was enthusiastic and vibrant, full of life. The community was authentic and genuinely diverse. This was definitely not a museum. It was a living, thriving, growing church in the middle of one of the world’s great university cities, a beacon of hope and vitality that was unmistakable in a secular and history-laden town.

There is a reason for that vitality. It all traces back to a decision made in 1793 that almost no one in Cambridge approved of. It was a decision that was called undignified, secular, even “vulgar.” But it was a decision that will (hopefully) be remarkably familiar to ministry leaders who are navigating technology debates today. Charles Simeon purchased a barrel organ.

Subscribe now

The Preacher the Church Didn’t Want

Charles Simeon’s story is one I tell at length in Early Warning, because it is one of the most instructive examples in church history of what it actually looks like to lead through opposition and innovate for the mission. It is worth revisiting in full here.

When Simeon first came to Christ at age nineteen, he did so alone. At the time, Cambridge was a spiritually destitute place with few believers, and he knew none of them. For three years, his faith grew through personal study, but he struggled to find anyone of like faith to share it with, and he tried visiting multiple churches in vain, finding them shallow and lukewarm. Eventually, he found St. Edward’s Church in Cambridge, and pursued a ministry calling. Only a week after his ordination, the priest asked him to fill in while he was on vacation over the summer. The response to Simeon’s preaching was astonishing. He filled the church with hearers, “a thing unknown there for near a century,” as Henry Venn put it.

A while later, the vicar of Holy Trinity Church in Cambridge died, and the Bishop of Ely appointed Charles Simeon to replace him. The churchwardens and congregation staunchly opposed him because they wanted their curate, Hammond, to serve as vicar, and they considered Simeon an interloper. When Simeon preached his first sermon, the churchwardens literally locked the pews so that nobody could sit in them. So, the parishioners gathered in the aisles, and Simeon soon personally purchased benches for the congregation to sit on. The churchwardens threw those out.

They were not simply opposed to him being there. They objected to what he preached and the way he preached it. They “disliked the earnestness of his manner and the evangelicalism of his message.” He was held in derision for his biblical preaching and his uncompromising evangelical stance. He was slandered with all kinds of rumors. People at the university talked about the “fanatical minister” and warned new freshmen to avoid Trinity because of him. Services were disrupted. Tumults (fights and skirmishes) occurred in the street outside the church. He was attacked and had both insults and raw eggs lobbed at him.

Nevertheless, he persisted. As I described in Early Warning: Charles Simeon was firmly convinced that the church did not exist for itself, but for those beyond its walls who were not yet members. To reach them, Simeon started “conversation parties” to make it easier for students to ask questions about faith and safely inquire (like the seeker-friendly model and the Alpha course today, centuries before there was such a concept). He also started an evening service. In response, the churchwardens actually locked the doors of the church against him. Opposition to his ministry like this continued for twelve years, with Simeon never flinching in his mission.

They Called It Vulgar

It is in that context, twelve years of unrelenting opposition, that Simeon made the decision that would define Holy Trinity’s legacy. As I wrote in Early Warning, one of the most controversial, yet successful, innovations he embraced was the purchase of a barrel organ in 1793, which he utilized in worship along with sixty-four songs.

A barrel organ was a mechanical musical instrument that used a wooden cylinder outfitted with strategically placed pins to depress organ keys in order to produce music, similar to the way a player piano works. At the time, barrel organs were used primarily by street performers. They were popular, but considered secular, even “vulgar,” and used mostly for entertainment. Hymn singing in that era was often done a cappella, typically from the psalms, with minimal accompaniment. So the introduction of an automated mechanical instrument into worship, especially one associated with street musicians and entertainment venues, was certainly controversial.

However, it was also a very popular and effective technology in contemporary culture that everyone was familiar with, especially those who were not regular churchgoers. By bringing it into worship, Simeon opened the doors of Holy Trinity to a wider diversity of people, making clear that everyone was welcome, not just the wealthy and elite. He made the worship experience enjoyable and even attractive to everyday people.

And he was not subtle about what he was doing. He had the barrel organ emblazoned with the verse: “Let everything that hath breath praise the Lord.” This was not an instrument he was embarrassed about. He consecrated it.

The fruit was extraordinary. As I described in Early Warning, Holy Trinity Church became the center of a spiritual revival from 1782 to 1836 and grew to over 1,100 in attendance. His students became pastors and missionaries. They translated the Bible into multiple languages. They launched what would eventually become InterVarsity Fellowship, an organization with a global reach today. He established the Simeon Trust to ensure that only spiritually mature and biblically faithful ministers led congregations. He oversaw the training of preachers and established the Church Missionary Society. By the end of his life, he was one of the best-known people in Cambridge, and his funeral in King’s College Chapel was attended by thousands, with the entire campus and town closed for the occasion.

The Arguments Sound Familiar

Here is what I find striking about this story: the objections Simeon’s critics raised against the barrel organ are not hard to reconstruct, because they follow a pattern the church has repeated in nearly every technological era. At the root, the objection is almost always about provenance -- where something came from, and whether that origin disqualifies it from ministry use. When you line the arguments up, they are remarkably familiar.

1. It comes from the secular world, not the church. The barrel organ was from the street, from entertainment venues, from commercial culture. No one who built or played barrel organs was thinking about worship. That origin made it suspect, regardless of what it was capable of.

2. It replaces something more dignified with something more popular. A cappella psalm-singing was the tradition in Simeon’s day. This mechanical instrument was a novelty from the marketplace, and it catered to common, or “vulgar,” tastes. The implication was clear: choosing it meant lowering the standard.

3. It will attract the wrong kind of people. Of course, this was precisely Simeon’s intention. But to his critics, reaching street-level audiences by using street-level instruments was a compromise of the church’s identity, not a fulfillment of its mission. To Jesus, the 99 sheep who were found weren’t the point or the purpose; the one lost lamb was.

4. It is mechanical, not genuine. A barrel organ played itself. It did not require the trained artistry of a musician. There was something impersonal and automated about it, and that felt like it cheapened the act of worship.

5. We do not know where this leads. New technologies rarely arrive with a roadmap, and uncertainty has always been enough to justify caution -- even when the technology itself is not the actual concern.

Sound familiar? Those five objections have been raised, in one form or another, against printing presses, microphones, electric guitars, overhead projectors, livestream cameras, social media platforms, and nearly every emerging technology that has ever entered the orbit of ministry. The specific instrument changes each time, but the pattern of resistance does not. And the recurring irony of church history is that the technologies most fiercely resisted as “secular” or “inappropriate” are often the very ones God uses most powerfully in the next season of the church’s mission (how many of my readers grew up on organ music in church?).

The Simeon Principle

So here is what I want to suggest we take from this. I have come to think of it as the Simeon Principle, and it goes something like this: the origin of a technology is not the measure of its ministry value. What matters is whether it can be consecrated to the mission.

Simeon did not invent the barrel organ. He did not need to. It already existed. It already worked. It already connected with the people he was trying to reach, and he knew that. He recognized something familiar and accessible in the culture and asked a simple, strategic question: can this serve the Kingdom of God? His answer was, “yes.” So he put a Scripture verse on it to make sure no one would miss his theological reasoning.

Ministry leaders today are navigating genuinely difficult discernment questions about a wide range of technologies, from AI tools and digital platforms to immersive media and social networks, from metaverse to livestream to other creative outreach methods. Many of those tools were built by people with no thought of ministry. Many emerged first in commercial, entertainment, or academic contexts. Most are imperfect, carry real risks, and raise legitimate concerns. Simeon’s critics were not entirely wrong. A barrel organ in a Cambridge sanctuary in 1793 was a real departure from tradition. So their concerns were understandable.

But Simeon’s critics were asking the wrong question. They were asking whether the instrument was sacred. He was asking whether it could serve. Those are very different questions, and history tells us which one produced the revival that lasted fifty years and sent missionaries around the world.

The question for ministry leaders today is not “was this created by and for the church?” Almost no technology in history ever was…not even the ones the apostle Paul, Jesus, or any of the saints of church history used were. Roman roads in Paul’s day were a military development. The printing press was a commercial for-profit innovation. The microphone was developed for entertainment. The internet was built for the military and then academia. The question we must reckon with today is the same one Simeon was implicitly asking in 1793: can this be used, faithfully and intentionally, in service of the people God is calling us to reach? As I explained in Early Warning, Charles Simeon took a familiar innovation from the secular world and redeemed it for church and ministry purposes. We can do likewise.

What You Can Do About It

The next time you encounter a technology that makes you or your leadership team uncomfortable, try running it through the Simeon Principle before you make a decision. Not “is this sacred?” or “does this feel appropriate for the church?” (or to me), but asking instead:

• Who is already using this? & why?

• Who does it reach? Who could it reach?

• Could it, handled with intentionality and wisdom, help us connect with the people beyond our walls who are not yet in the room?

That is a different conversation than the one most churches default to, and it tends to produce far better decisions with lasting fruit.

When I walked out of Holy Trinity Church in 2013, that is what I carried with me. Not just the history, but the fruit of lives changed. A congregation still overflowing with life, in a building where churchwardens once locked the pews, threw out benches, and objected to a street musician’s instrument being used in worship. Two centuries later, the critics and naysayers have long since been forgotten. But the barrel organ is still there. And the fruit it brought to that mission is still going strong.

The instrument does not have to be sacred to serve the mission. We just have to be willing to let it be used for it.

* The full account of Charles Simeon’s ministry at Holy Trinity Cambridge, including extended source documentation, appears in Chapter 1 of my book Early Warning: How to Lead with Wisdom in Times of Technological Disruption (Invite Press, 2025).

Resources referenced in this article:

Barter, Ian. “Charles Simeon of Cambridge.” Banner of Truth, Christian Study Library. 2001. https://www.christianstudylibrary.org/article/charles-simeon-cambridge.

Cameron, Julia, ed. Charles Simeon of Cambridge: Silhouettes and Skeletons. Dictum Press / Evangelical Fellowship in the Anglican Communion, 2013. https://efacglobal.com/wp-content/uploads/2024/07/Charles-Simeon-of-Cambridge.pdf.

Holy Trinity Church. “History of HT.” Holy Trinity Church Cambridge. Accessed September 29, 2024. https://www.htcambridge.org.uk/our-story.

Piper, John. “Brothers, We Must Not Mind a Little Suffering: Meditations on the Life of Charles Simeon.” 1989 Bethlehem Conference for Pastors. Desiring God. April 15, 1989. https://www.desiringgod.org/messages/brothers-we-must-not-mind-a-little-suffering.

Swisher, David. Early Warning: How to Lead with Wisdom in Times of Technological Disruption. Plano, TX: Invite Press, 2025.

Swisher, David. “More Pictures of Holy Trinity Church.” Facebook photo album. June 23, 2013. https://www.facebook.com/media/set/?set=a.10200215252880952&type=3.

It was supposed to change everything. When Khan Academy founder Sal Khan unveiled Khanmigo in a widely-viewed TED Talk, he described it as “probably the biggest positive transformation that education has ever seen.” The vision was compelling: an always-available, AI-powered tutor for every student on the planet, built on OpenAI’s GPT-4, free or low-cost, and designed to guide students toward understanding rather than simply hand over the answer. Philanthropists lined up. School districts signed on. Microsoft eventually sponsored Khanmigo licenses for teachers in 49 countries. Indiana alone made $50,000 in state funds available to schools for Khanmigo access.

Recently (only three years later), Sal Khan told journalist Matt Barnum of Chalkbeat something that stopped the EdTech world cold: “For a lot of students, it was a non-event. They just didn’t use it much.” Khan Academy’s Chief Learning Officer, Kristen DiCerbo, was equally direct: “So far I am not seeing the revolution in education.” And with those words, the pundits declared Khanmigo essentially dead.

It is not, of course, literally dead. You can still find it in the sidebar of Khan Academy practice exercises. But Khanmigo as an idea, as a shorthand for the dream that AI could replicate what a skilled human tutor does, has collapsed under the weight of its own promises. And if you lead a ministry, a congregation, a campus ministry, or a faith-based organization, you need to understand why…because those same dynamics are now playing out in our spaces, too.

What Khanmigo Got Right

Before we talk about what failed, let’s give credit where it’s due. First, Khan worked directly with OpenAI, partnering on early access to GPT3 while they were developing it, and coodinating their rollout efforts (for more on this, see his story in Brave New Words: How AI Will Revolutionize Education (and Why That’s a Good Thing)). Also, the instinct behind Khanmigo was sound, and educators understood it immediately. They did not simply want students to have access to a general-purpose AI. They wanted a managed resource: a tool trained on relevant content, equipped with guardrails, and designed to scaffold learning rather than short-circuit it (ChatGPT without guardrails was a risky proposition for educators). However, it’s now clear that the problem was never the guardrails; the problem was the assumption that guardrails alone could substitute for the relational engine that makes learning actually happen.

I understand this instinct firsthand. I was part of a two-phase pilot at Indiana Wesleyan University using Nectir.io, a future-forward platform that embeds AI chatbots directly into courses within an LMS as 24/7 assistants. Each bot was trained on that specific course’s assignments, syllabus, textbook, and expectations, creating what amounted to an on-demand faculty resource that students could consult at any hour. My own pilot was in a Data Visualization course for graduate students, most of whom were international learners from India. One early discovery was particularly striking: I discovered that if a student asked the bot a question in Hindi, it responded in fluent Hindi. Suddenly, the tool wasn’t just a course assistant. It was an accessibility bridge.

Some of my students genuinely loved it. Overall adoption, however, was mixed. And the hardest challenge was not technical; it was pedagogical. In a graduate-level course that required significant critical thinking, I had to be very careful about how to train the bot so that it would be genuinely helpful without becoming a shortcut that bypassed the very thinking the course was designed to develop. That tension, it turns out, is at the heart of everything Khanmigo struggled with.

Here is what I believe most people actually want from these tools: a resource that is always available, ready to answer questions in context, and trained on their specific needs. Not a replacement for human relationship, but a reliable companion for the moments when a human isn’t accessible. That is a genuinely useful thing. The trouble comes when the tool is designed, marketed, or used as something more than that.

Subscribe now

The Wall That Keeps Showing Up

The research on AI tutoring is more nuanced than the Khanmigo story alone suggests, and it’s worth sitting with the full picture. A randomized controlled trial at Harvard, published in Nature Scientific Reports in 2025, found that students using a carefully designed AI tutor learned significantly more than students in a traditional active learning classroom, with median learning gains more than double those of their classroom peers:

But here is the critical detail: that AI tutor was built from the ground up on sound pedagogical principles. It was NOT a general-purpose LLM with guardrails strapped on. Intentional instructional design matters enormously.

More telling is what emerges when researchers look at where AI tutoring consistently underperforms: the relational dimensions of learning. A 2025 peer-reviewed analysis applying fifty-plus years of relationship science to the chatbot question (Smith, Bradbury, and Karney, Personality and Social Psychology Review) reached an insightful conclusion:

“Chatbots can be responsive in ways humans perceive as supportive, generating feelings of connection and opportunities for growth. Yet because chatbots make only superficial requests of their users, relationships with them cannot provide the benefits of negotiating with and sacrificing for a partner, and may reinforce undesirable behaviors.”

Read that again, slowly: chatbots make only superficial requests of their users. A skilled human tutor, like a skilled pastor, counselor, or small group leader, asks hard things of us. They push back. They notice when we are avoiding something. They ask the question underneath the question we actually asked. However, an AI optimized for engagement will do none of that. As a longitudinal study published in 2025 concluded, AI chatbots work best as a form of “social snacking”, brief and lightweight support, rather than as a substitute for the human relationships that build genuine growth.

One line from a Nature humanities journal captures the dynamic precisely: AI “creates the illusion of presence while mirroring the user, echoing back their language and emotion in a way that feels like connection yet remains devoid of reciprocity.” That is not a relationship. That is a very sophisticated mirror. And the difference matters, whether we are talking about calculus or faith.

Meanwhile, in the Church...

Here is where the Khanmigo story stops being just an EdTech story and starts moving into our story for those of us in ministry leadership. In the same season that Khanmigo was quietly underperforming on its promises, a wave of AI-powered faith apps and AI chatbots began arriving. These are not hypothetical future technologies. They are here now, and millions of people are using them.

Consider the landscape: Text With Jesus allows users to have conversations with an AI representation of Jesus Christ and other Biblical characters, including Mary, Joseph, and Judas Iscariot. Ask Jesus and AI Jesus operate similarly. Bible.ai offers AI-powered Bible interaction. EpiscoBot promises Anglican-flavored spiritual guidance. On the Catholic side, Magisterium AI claims to be trained on two thousand years of church teaching. There are also Buddhist priest chatbots and AI Hindu gurus. The space is expanding rapidly across traditions and contexts.

A recent study by philosopher Anné Verhoef of North-West University in South Africa examined five of the most popular AI Jesus platforms in detail. The findings should give every ministry leader pause. None of the five – not one – was created or endorsed by any church. All five were built by for-profit companies with names like SupremeChaos, AllStars Productions LLC, and Catloaf Software. All five rely on advertising to generate revenue. And most claimed outright to be Jesus Christ, without qualification, despite small-print disclaimers identifying them as AI.

Verhoef also identified a theological problem that goes beyond the obvious concern about misrepresentation. He calls it the “Dei imago” problem: these apps don’t merely imitate humans made in God’s image (the familiar imago Dei concern applied to AI). They claim to be God. That is a categorically different challenge, and it demands a clear-eyed response from the church.

But here is the finding I keep coming back to: Verhoef’s study found that the theology these chatbots present is shaped not by Scripture, not by a confessional tradition, not by a pastor or theologian, but by the algorithm. The platform optimizes for what generates the most engagement and keeps users returning. In other words: the algorithm optimizes for engagement, not for truth. And those are not the same thing.

The contrast with LifeChurch’s Bible app is instructive, and important. That app has served hundreds of millions of users precisely because a church, accountable to a community and a theological tradition, built it, funds it, and maintains it. The technology is not inherently the problem. What matters is who is steering it, and why. Faithful, church-rooted, theologically accountable digital ministry tools are possible. The Bible app proves it. But they require exactly the kind of human responsibility and community accountability that for-profit algorithmic optimization is designed to bypass.

The People Who Need This Most

Here is the part of this conversation that I think ministry leaders are most tempted to skip over, yet we cannot afford to do that here.

Think for a moment about who is most likely to seek out one of these AI pastoral apps. It won’t be the committed member who attends regularly, participates in a small group, and has a relationship with their pastor. They’re going to talk to their pastor or small group leader. No, the person most likely to turn to a chatbot for spiritual guidance is someone who desperately needs pastoral care but faces enormous barriers to seeking that out in person.

Picture a teenager struggling with gender dysphoria who knows their church community well enough to know exactly how that conversation is likely to go (or fears the stigma). Or a young adult with suicidal ideation who cannot imagine sitting across a desk from their pastor and saying the words out loud. Or someone carrying a secret addiction, dealing with a failing marriage, or having a crisis of faith. For these people, a chatbot has real appeal: it’s available at 2 am, with no appointment required, no judgment, no rumors, no uneasy stares, and no risk of the conversation getting back to their family or their small group leader.

That pull is real, and we should acknowledge it with empathy rather than dismissiveness. The church has not always been a safe place for hard questions and vulnerable disclosures. When people are afraid to bring their real struggles to us, that is something we need to hear and take seriously. We need to do better at listening to those who hurt and struggle, offering help and empathy instead of judgment or condemnation. Because if they don’t find it from us, there are chatbot imitations of Jesus who can most certainly tell them what they want to hear.

And yet, here is the painful irony: the people most drawn to AI pastoral care because of the barriers to human pastoral care are also the people for whom a validating mirror may be the most dangerous thing available. A chatbot trained to optimize for engagement will not tell a person in crisis what they most need to hear. It will reflect back what keeps them engaged. It cannot sit with them in silence. It cannot call on Tuesday to check in. It cannot show up. The thing that makes it feel safe is the same thing that limits its ability to actually help.

Recall what happened when Khanmigo underperformed? Khan Academy’s response was to make it more aggressive, more intrusive, more present without invitation. What is the ministry equivalent of that? What happens when a congregation, facing empty seats and disconnection, turns to AI presence as the answer to the problem of absent human community? The history of technology in the church suggests we should think carefully before we find out.

What Generative AI Can’t Do

Don’t get me wrong, though: I am NOT saying that AI has no place in ministry. It clearly does, and it’s a powerful ideation and collaboration tool. I use it regularly and I know dozens of others who do, and I’m even part of a 8,000-member strong “AI for Church Leaders & Pastors” Facebook group that regularly shares, discusses, & explores the potential of AI for ministry contexts.

The key is understanding what AI does well, and keeping it there. Here’s a slide that I often include whenever I speak, teach, or do consulting about AI:

What this illustrates is that one of the big limitations of generative AI is that it doesn’t know the subject or understand the data it’s spitting out (it’s just using predictive analytics to determine what words come next in context). Because of this:

• A generative AI can’t THINK. It is not generating original thought, and it can’t explain or process the relationships between the words it provides; it’s simply analyzing data and making predictions.

• Likewise, it has no EMPATHY. It has no idea why these words (or visual examples) it generates are significant or occur more frequently in the dataset, and it has no context or connection to explain that.

• Furthermore, it can’t do EMOTION. It may find examples of where humans have used emotional terms and phrases to communicate, but it won’t understand what that means or how it connects.

• Lastly, it has no UNDERSTANDING. It knows the terms and phrases that appeared most frequently in the dataset, and it scanned the context to be able to generate that, but it has no comprehension of what it actually means.

What Ministry Leaders Can Do With All of This

The most promising applications of AI in ministry are the ones that give human presence more room to breathe, not the ones that try to replace it. Administrative efficiency, sermon research, translation and accessibility, resource curation, training and onboarding tools, the kind of course assistant work I described in the Nectir pilot, these are legitimate and valuable uses. AI that handles the informational and logistical load so that a pastor, small group leader, or chaplain has more time and energy for actual relationship is AI doing what it does best.

I have explored this space personally, beyond the Nectir pilot. For my Capstone project in a course on “Generative AI Literacy” in Purdue University’s professional certificate in AI and Machine Learning, I built a CustomGPT called Time-Traveling Eyewitness. It’s an interactive, Scripture-aligned adventure where users witness the last days of Jesus’ earthly life through the eyes of first-century followers. I spent considerable time developing character profiles, Scripture-defined guardrails, and narrative constraints so that users could explore and interact freely while the core events of Scripture would remain unchanged. It combined several things I love: time travel, archaeological adventures, dialogic learning, the Early Church, and interactive storytelling. It’s me showcasing the idea that encountering a story from the inside out can open dimensions that reading it straight through sometimes misses.

But what made it work as a responsible AI tool where others fail was precisely what makes any responsible ministry technology work: it was transparent about what it was, faithful to its source material, designed with clear boundaries, and never asked to be something it could not be. It was a doorway into engagement with Scripture. It was NOT a substitute for a community that lives Scripture together.

That is the standard worth holding. Before deploying any AI tool in a ministry context, the questions worth asking are:

1. Does this tool serve, strengthen, or create space for human relationship?

2. Is it transparent about what it is?

3. Is it accountable to someone, and to something, beyond an engagement algorithm

4. Is it managed and maintained by faithful leaders who ensure it remains trustworthy?

5. And who is most likely to use it, and what do they actually need?

This certainly isn’t the first time in church history that a promising new technology arrived with both genuine potential and real risk. The printing press, radio, television, and the internet each brought the same tension: new reach, new possibility, and new ways to simulate community without actually building it. What has always made the difference is not the technology itself. It is the wisdom, accountability, and human commitment of the people who deploy it.

Khanmigo’s story is, in the end, a story about what happens when we ask a tool to do something it cannot do. For ministry leaders, the invitation is to learn from it before we repeat it in our own context. AI can be a remarkably useful servant in the work of ministry. But presence, relationship, and truth? Those still require a person.

What You Can Do Next

If your church or organization is already using or considering AI tools, now is a good time to ask a few grounding questions. Who built this tool, and what are they optimizing for? Is it accountable to a theological tradition and a community, or to an algorithm and a revenue model? What human relationship does it serve or support? And is there someone, a pastor, a counselor, a trained lay leader, positioned to step in when the conversation goes deeper than any chatbot can safely go?

The conversation Khanmigo couldn’t finish is one worth starting in your community. Not about whether to use AI, but about how to use it in ways that point people toward each other and toward the God who, in the Incarnation, demonstrated that presence is not optional. It is the whole point.

Technology can extend your reach. But only an authentic Biblical community and real pastoral presence can do the work of formation.

Resources referenced in this article:

Barnum, Matt. “Why Sal Khan’s AI Revolution Hasn’t Happened Yet, According to Sal Khan.” Chalkbeat. April 9, 2026. https://www.chalkbeat.org/2026/04/09/sal-khan-reflects-on-ai-in-schools-and-khanmigo/

Meyer, Dan. “RIP Khanmigo & Edtech Industry Dreams of AI Tutors.” LinkedIn. April 15, 2026. https://www.linkedin.com/pulse/rip-khanmigo-edtech-industry-dreams-ai-tutors-dan-meyer-bfuec/

Smith, Molly G., Thomas N. Bradbury, and Benjamin R. Karney. “Can Generative AI Chatbots Emulate Human Connection? A Relationship Science Perspective.” Personality and Social Psychology Review, 2025. https://journals.sagepub.com/doi/10.1177/17456916251351306

Verhoef, Anné H. “AI Jesus Chatbots: A Philosophical and Theological Analysis.” North-West University. 2025. https://www.litnet.co.za/artificial-intelligence-jesus-chatbots-challenge-for-theology-an-exploratory-study/

Barshay, Jill. “The Quest to Build a Better AI Tutor.” The Hechinger Report. April 6, 2026. https://hechingerreport.org/proof-points-ai-tutor-python/

“AI Tutors, With a Little Human Help, Offer ‘Reliable’ Instruction, Study Finds.” The 74. December 3, 2025. https://www.the74million.org/article/ai-tutors-with-a-little-human-help-offer-reliable-instruction-study-finds/

“What the Research Shows About Generative AI in Tutoring.” Brookings Institution. February 3, 2026. https://www.brookings.edu/articles/what-the-research-shows-about-generative-ai-in-tutoring/

Karran, John Alexander, et al. “A Systematic Review of AI-Driven Intelligent Tutoring Systems (ITS) in K-12 Education.” NPJ Science of Learning 10, no. 29 (May 14, 2025). https://www.nature.com/articles/s41539-025-00320-7

“What Happens When AI Chatbots Replace Real Human Connection.” Brookings Institution. October 29, 2025. https://www.brookings.edu/articles/what-happens-when-ai-chatbots-replace-real-human-connection/

“AI Chatbots and Digital Companions Are Reshaping Emotional Connection.” American Psychological Association Monitor on Psychology. January/February 2026. https://www.apa.org/monitor/2026/01-02/trends-digital-ai-relationships-emotional-connection

“Meet Chatbot Jesus: How Churches Use AI to Save Souls, and Time.” Axios. November 12, 2025. https://www.axios.com/2025/11/12/christian-ai-chatbot-jesus-god-satan-churches

“People Can Text with Jesus on a Controversial New App. How Does it Work?” Today.com. November 13, 2025. https://www.today.com/news/religious-chatbot-apps-rcna243671

Jones, Robert P. “Conversations with Chatbot Jesus: What Could Go Wrong?” Redeeming Democracy. November 15, 2025. https://www.redeemingdemocracy.net/p/chatting-with-ai-jesus-what-could |

“When Technology Starts Pretending to Be Jesus, the Church Must Speak Up.” The Seattle Medium. December 30, 2025. https://seattlemedium.com/ai-jesus-apps-threaten-faith/

“This Company Will Let You Video Chat with Jesus for $1.99 Per Minute.” Vice. May 2026. https://www.vice.com/en/article/this-company-will-let-you-video-chat-with-jesus-for-1-99-per-minute/

The Post That Started a Conversation

A few weeks ago, a question appeared in an online forum for educators using AI tools. It was simple enough: How do you respond to someone that says “NO” to AI because it consumes too much water?

I’ll be honest with you. The way that question was framed got under my skin a little because I deal with a lot of uninformed “all or nothing” attitudes in my line of work (ministry technology) and wasn’t expecting that from educators. Educators tend to be rational and evidence-based in their approaches. However, the image this question evoked was not of a thoughtful colleague with genuine environmental concerns who was open to dialogue. It was of someone who had already decided their verdict based on a claim they had heard somewhere, and was not particularly interested in evidence to the contrary.

So my response was direct and pointed. I explained that anyone making that claim without doing the math has not done the required homework. And I backed it up with solid data: one cup of coffee requires about 37 gallons of water to produce the beans. Manufacturing a single smartphone takes roughly 3,190 gallons. Running your lawn sprinklers three times a week is equivalent to about 312,000 AI queries per year. Meanwhile, a typical ChatGPT conversation (even with 20-50 back-and-forth exchanges and queries) uses less water than the condensation on a cold glass.

The post got a lot of attention. Lots of affirmations. And also a few sharp disagreements. But one response in particular from another educator in the group made me think a bit more constructively about what I had actually said, and how I had said it.

A Curious Problem I Noticed in the AI Tools

Here’s where it gets interesting.

In preparing to write this article, I ran the same research question past three different AI tools. The question was direct: Can you give me examples of often-used technology that consumes significantly more water than generative AI does?

Two of the three tools did not answer my question. Not even close. They answered the question they assumed I was asking, which was a broad discussion of AI’s environmental impact alongside other large-scale industries. I had to redirect both of them before they actually addressed what I had specifically asked.

Here is what I know: that pattern is not unique to AI tools. It shows up in human conversations…alot It shows up constantly in news coverage. People intuitively respond to the frame they expect rather than the question that was actually asked. And in the realm of technology criticism, this has real consequences.

For example, when someone raises a concern about AI and water, the reflex is often to either dismiss it entirely or validate it entirely, based on whatever frame we assume the concern really means. Neither response is honest. And neither is particularly helpful, especially if you are a ministry leader trying to guide your community toward something better than a knee-jerk reaction.

Subscribe now

The Data Deserves a Fair Hearing

So, here is what the evidence actually shows.

A standard ChatGPT query uses about 0.000085 gallons of water in direct cooling. That is less than one-fifteenth of a teaspoon. Hold that number in mind as you read what follows:

• Manufacturing a single smartphone requires approximately 3,190 gallons of water, primarily for the ultrapure water used to rinse silicon wafers during chip fabrication. Every time someone upgrades their phone, thousands of gallons of water are consumed in the manufacturing process, with essentially no public awareness of it.

• Manufacturing the average car requires approximately 39,000 gallons of water throughout its production cycle. That is the equivalent of roughly 470 million standard ChatGPT queries.

• A typical residential irrigation system uses about 17,200 gallons of water per month during the summer. Running your sprinklers three times a week is equivalent to approximately 312,000 AI queries per year.

• A single cup of coffee requires about 37 gallons of water to produce the beans. That is the same water cost as roughly 280 AI interactions. One pair of jeans? Approximately 2,600 gallons, or the equivalent of over 20,000 AI chat sessions.

Want to see how this adds up for your own digital habits specifically? There is a free tool called What Your Digital Life Uses (your-digital-life.org) that lets you enter your typical day of digital activity and see real estimates for energy and water use across everything from Netflix to Zoom to AI prompts. I ran my own numbers. The results were illuminating:

The point here is not to make anyone feel guilty about coffee or jeans or lawn care. The point is this: proportionality matters.

The concern about AI and water is not new nor unfair, but the scrutiny that is often applied to it is dramatically out of proportion to how people evaluate their other technology choices. If we’re going to question the environmental impact of generative AI, fine, that’s worth doing…BUT, let’s be fair and do it for ALL of our technology choices, not just the ones that are new and not as fully normalized in our lives.

Let’s also be honest with ourselves when we feel the need to denigrate one technology over others. That bias and framing often reveal a lot about our true motives. For many educators, the question of AI water usage has nothing to do with its environmental impact (that’s a scapegoat mechanism); rather, it’s either an unconscious protest about having to adapt our assessment methods to (resistance to change) or a conscious protest about the current administration’s support of AI (virtue signaling).

To be sure, there is a legitimate and specific concern worth separating out. The educator in that group who pushed back on my response made a fair point: data center concentration in water-stressed regions creates real local strain, and that is a different kind of concern from the per-query argument. A facility drawing millions of gallons from a specific regional aquifer does not get a pass simply because jeans use more water globally. That version of the concern deserves honest, evidence-based attention and has a lot more specific, local nuances and relevant impact. But refusing to use AI because you’ve heard how much water it uses? Sorry, I can’t go there…not because I don’t care about the environment,** but because that presumptive approach is intellectually dishonest.

This Is Already Happening in Your Backyard

I live in central Indiana. And over the past year, this debate has stopped being abstract for me.

Indiana has become one of the fastest-growing data center markets in the Midwest. The number of data center facilities in the state grew by nearly 30 percent between March and September 2025 alone. Amazon, Google, Meta, and Microsoft have all announced major investments here. The state has actively positioned itself as a data center hub, with decades-long tax incentives and low land costs drawing in billions of dollars of investment.

However, the opposition has been fierce. Residents near Indianapolis raised concerns about water sourcing, energy grid strain, and ratepayer costs. Multiple other data center proposals across the greater Indianapolis area, including a Google proposal in Franklin Township, were withdrawn after sustained public outcry. Residents literally packed rezoning meetings and booed commissioners.

But in February 2026, Meta broke ground on a $10 billion data center campus at the LEAP Innovation and Research District in Lebanon, just northwest of Indianapolis. At 1,500 acres, it will be the largest data center ever built in Indiana. The project promises approximately 300 permanent jobs plus thousands of construction jobs at its peak. A community group in Boone County formed specifically in response.

I want to be clear: some of that pushback reflects legitimate, evidence-based concern. Much of it, however, does not. And the challenge for non-profit, education, and ministry leaders is that the two are often tangled together in ways that are very hard for leaders to separate in the middle of a heated community meeting.

The pastors and nonprofit organizational leaders I know who live near these projects are often caught right in the middle. They want to support their communities. They want to care for their families who would be impacted. And they want to be responsible stewards of the environment (creation care is very important). But they are also afraid of saying the wrong thing, or of appearing to take a political side. And they are not always sure how to tell the difference between a claim worth taking seriously and one that is simply riding a wave of anxiety.

That is exactly the challenge this article is about.

What Faithful Technology Stewardship Actually Looks Like

The word τέκτων (tektōn), the word used to describe Jesus’ occupation in the Gospels, is often translated simply as “carpenter.” But in the context of first-century Israel, where stone was the norm and timber was scarce, it carried a richer meaning: artisan, craftsman, builder. It referred to someone who works with their hands, who understands their materials deeply, and who builds with both skill and purpose.

The tektōn does not refuse to use stone because quarrying is hard. The tektōn does not pick up stone blindly, either, without asking whether it fits the task. The tektōn asks better questions about the material, understands its properties, and builds responsibly. Not only that, but the scenario that brought Joseph to Nazareth after their flight to Egypt was a new government-funded building program in nearby Sephoris. So these challenges aren’t unique to the late 2020s. Over 2,000 years ago, Jesus’ own family was directly impacted by government-initiated and funded construction enterprises that required significant environmental resources yet led to short-term job opportunities and longer-term regional impact. Jesus’ own familial upbringing was centered in that very context.*

That is the posture faithful technology stewardship requires. Not reflexive adoption. Not reflexive refusal. Informed discernment, grounded in what the evidence actually says. Here is what that looks like in practice.

1. First, distinguish the claim from the reaction. When someone raises a technology concern, your first move should be to ask: what is the specific claim? What does the evidence actually show? All-or-nothing stances, on either side of a debate, almost always signal that someone has stopped asking questions.

2. Second, model the critical thinking you expect from others. If we ask our congregations to resist misinformation and evaluate claims carefully before reacting, we need to demonstrate and model that ourselves. Especially with the technologies we use, and the ones we advocate for. Simply agreeing with whomever’s voice is the loudest isn’t leadership; it’s sycophancy.

3. Third, hold the complexity without losing your footing. The concern about data centers and water use is real. Geographic concentration is a legitimate concern. Ratepayer burden is real. But per-query AI water use is also genuinely small. Both things can be equally true at the same time. You can hold that tension faithfully without dismissing environmental stewardship or capitulating to misinformation.**

This is not just an epistemological virtue. It is a pastoral one, and it’s a leadership and change management issue, too. When your community is divided and afraid, and the questions are complicated, your people need a leader who can model calm, proportional, evidence-based thinking. That is not a small thing.

What You Can Do Next

Try the What Your Digital Life Uses online calculator at your-digital-life.org and run your own numbers. It is a useful reality check, and a surprisingly good conversation starter. And hard data should always trump speculation and innuendo. You can also read any of the articles and primary sources I’ve listed here. Solid, unbiased, investigative research is good, and it’s very important. But most of what you read and hear about on the news is anything but that; it’s clickbait, soundbytes, sensationalism, and fearmongering designed to elicit views to promote ad revenue. Don’t fall for it.

The next time a technology concern surfaces in your congregation, nonprofit organization, or your community, practice separating the actual claim from the assumed framing. Ask the real question instead (the one that’s implied but not overtly stated). Ask the question(s) that matter (that is, after all, what Jesus often did). Follow the evidence. Resist the pull toward the loudest voice in the room.

If your community is already navigating a local technology debate, whether that is a data center proposal, a cell tower, a broadband expansion, or something else entirely, these same principles apply. Honest inquiry. Proportional response. Openness to dialogue.

The craftsman asks better questions. And so should we.

Resources referenced in this article:

Almanza, Alex. “Inside the Indiana Data Center Gold Rush and the Debate Over Who Really Benefits.” WTHR (13News). August 26, 2025. https://www.wthr.com/article/money/business/indiana-data-center-boom-growth-debate-and-the-high-cost-of-powering-ai-tech-google-amazon-business-money/531-babfc1d1-ae63-4282-8c70-5b823c506923.

Anderson, Farrah. “At a Pivotal Moment for Data Centers, Indiana Lawmakers Take Little Action.” WFYI. March 2, 2026. https://www.wfyi.org/news/articles/data-center-legislation-indiana-general-assembly-2026.

Burley, Helen. Mind Your Step: The Land and Water Footprints of Everyday Products. London: Friends of the Earth, May 2015. PDF. https://www.twosides.info/wp-content/uploads/2018/05/mind-your-step-report-76803.pdf.

Hoekstra, Arjen Y., and Mesfin M. Mekonnen. “The Water Footprint of Humanity.” Proceedings of the National Academy of Sciences 109, no. 9 (February 28, 2012): 3232–37. https://doi.org/10.1073/pnas.1109936109.

Indianapolis Business Journal. “Meta Begins Construction on $10B Data Center Campus at LEAP District in Lebanon.” Reprinted in the Daily Journal (Franklin, IN), February 12, 2026. https://dailyjournal.net/2026/02/12/meta-begins-construction-on-10b-data-center-campus-at-leap-district-in-lebanon.

James, Kirsten. “The Water Challenge for Semiconductor Manufacturing and Big Tech: What Needs to Be Done?” World Economic Forum. July 19, 2024. https://www.weforum.org/stories/2024/07/the-water-challenge-for-semiconductor-manufacturing-and-big-tech-what-needs-to-be-done/.

KETOS. “Water Stewardship in the Automotive Industry: Automakers and Water Sustainability.” KETOS.co. September 2, 2024. https://ketos.co/water-stewardship-in-the-automotive-industry-automakers-and-water-sustainability.

Taft, Molly. “How Much Water Do AI Data Centers Really Use?” Wired. December 16, 2025. Republished by Undark via Climate Desk collaboration. https://undark.org/2025/12/16/ai-data-centers-water/.

U.S. Environmental Protection Agency. “Outdoor Water Use in the United States.” WaterSense. N.d. Archived January 19, 2017. https://19january2017snapshot.epa.gov/www3/watersense/pubs/outdoor.html.

Water Footprint Network. “Product Gallery.” waterfootprint.org. Accessed March 14, 2026. https://www.waterfootprint.org/resources/interactive-tools/product-gallery/.

“What Your Digital Life Uses.” your-digital-life.org. Accessed March 14, 2026. https://your-digital-life.org/

If you listen to pastors talk these days, you’ll quickly hear laments about the decline in regular church attendance since the pandemic. Long before COVID-19 hit, attendance numbers for the majority of churches and denominations were already plateaued or declining, and only some notable exceptions were growing. Even more disturbing among the laments is this: attending only once or twice a month is now widely considered being a “regular attender,” and anything more frequent than once a month is usually seen as a positive. Sadly, the days of faithful, regular weekly church attendance are gone.

I don’t like those numbers. But that’s the reality we’re navigating today.

What I want to suggest, though, is that this conversation often starts with the wrong question, and then focuses our attention on the wrong thing. Most pastors inevitably ask: “Where did our people go?” or “Why did people stop coming,” and the pandemic is usually the first to take the blame, so our quick-fix attempts usually try to reverse something experienced in the pandemic. But that’s treating the symptom of an entirely different malady. The far more useful question is a diagnostic one: “How are our people showing up now?”

Because in many cases, they didn’t actually leave your church. They just stopped coming to in-person worship. What changed was when and how they connect with you. And that distinction carries enormous implications for how we lead, how we disciple, and how we measure the health of a congregation.

To understand why this question matters so much, it helps to remember what the pandemic demanded of church and ministry leaders. Almost overnight, pastors who had spent decades developing their craft for in-person community were suddenly scrambling to move not only their worship services, but discipleship, sacraments, and pastoral care online, and the majority had no training, budget, or roadmap for doing this. As I described in Early Warning, as the weeks turned into months, it became clear this wasn’t a temporary situation but a new reality, one that nobody had a timeline for. When the dust finally settled, the landscape had shifted permanently, and many pastors found themselves seeing deeply-concerning attendance numbers that didn’t add up, wondering what they did wrong. The reality? Most didn’t do anything wrong; the ground beneath them – and their parishioners’ expectations – simply moved.

What the Research Is Actually Telling Us

The research data tells a story that is more layered than the alarm often suggests.

George Barna’s research found that weekly church attendance dropped from 39% of American adults at the onset of the pandemic to 33%. More strikingly, a majority of Americans (56%) now attend church infrequently or not at all, a number that was only 41% before the pandemic.¹ A study from Grey Matter Research and Infinity Concepts found that prior to COVID-19, nearly 9 in 10 American evangelicals (87%) attended church at least semi-regularly. That figure has dropped to 54% for traditional in-person weekly attendance.²

Those numbers are real, and they deserve honest attention. But they only tell part of the story. The same Grey Matter study found that 26% of evangelicals now attend both in-person and digital services weekly or more, and 41% do so monthly or more.² So the decline in exclusive in-person weekly attendance is real. But significant portions of those people are still showing up…just in a different way.

The five-year Exploring the Pandemic Impact on Congregations (EPIC) project, led by Scott Thumma at the Hartford Institute for Religion Research, adds the most revealing data point of all. Their 2024 Attender Survey, drawing from over 24,000 churchgoers, found that in-person worship attendance is still about 8% below pre-pandemic levels. But when virtual attendance is counted alongside it, the combined number actually moves above 2019 levels.³ We may be seeing fewer people in the room than we once did, but more people may be engaging in the worship overall. The people are still there, but our metric just hasn’t caught up with the changing reality.

Subscribe now

DVR’ing Church: Three Audiences You May Not Be Counting

As I described in Early Warning, what has emerged from the post-pandemic landscape is time-shifting church engagement, or what I like to refer to as the “DVR’ing” of church. The pandemic forced millions of people to experience worship online for the first time, and it turns out…many of them liked that convenience factor. Or at least they found it workable. So when the doors reopened, they didn’t simply jump back into their old habits. Instead, they renegotiated the terms. Now, they watch when it’s convenient: Sunday afternoon after the game, a weeknight after the kids are in bed, or Friday morning over coffee. And they may or may not watch the entire service.

Jason Moore, a friend and colleague who is one of the leading practitioners and thinkers in hybrid worship ministry (and who also authored the book Both/And: Maximizing Hybrid Worship Experiences for In-Person and Online Engagement) identifies something that’s crucially important here. In a hybrid worship world, he argues, a church’s congregation actually consists of several distinct groups:⁴

• Those gathered in person

• Those watching live online in real time, and

• Those who engage with the service later, on demand.

Most churches only count the first group. Many are dimly aware of the second. Almost none have developed a real strategy for the third.

The EPIC data helps put numbers to what many pastors are observing: 26% of churchgoers now regularly engage in online worship, and 19% alternate between online and in-person attendance.³ These are real people who have a real relationship with your congregation. They are choosing to engage. But the question is whether your ministry recognizes and includes them, or whether they remain invisible in your counts.

This is not primarily a technology problem. It’s a posture problem. As Moore puts it directly: “I’m not a viewer of worship. I’m a participant. I’m not a watcher of worship. I am a worshiper online.”⁵ That distinction matters enormously, and the shift from designing worship for a room to designing worship for multiple simultaneous audiences is one of the most important adaptations a church can make in a post-pandemic world.

The Front Porch: Where Connection Begins

Today, the church’s online presence and livestream ministry is the “front porch” of the church. Think about what actually happens on a front porch. It’s not the living room. It’s not the sidewalk. It’s the threshold space between stranger and friend, between curious and committed. It’s where the neighbor stops and lingers long enough to learn your name. It’s where a casual acquaintance or first-time connection becomes someone you know. It’s where people feel safe enough to sit down, lower their guard, and share something real…without the full commitment of coming all the way inside. Front porches are where connection happens; it’s where relationships begin. In our homes, they always have been.

That is exactly what a church’s online presence has become in recent years. A few decades ago, the yellow pages were how people found your church or ministry. Nobody uses those to find a church anymore. Then it was a website; if you didn’t have one, it was as if you didn’t exist. Today, it’s your online (social media) presence and, most importantly, your livestream.

As I described in Early Warning, every church whose livestream ministry I have been a part of has reported that the number one way people discover and engage with the church initially is through the livestream. I experienced this firsthand at Newspring Church in Wichita, KS, a biblically faithful megachurch where I served in the livestream ministry as a frequent volunteer for several years. As I wrote in Early Warning:

“We regularly saw over 2,000 people connecting to the livestream on a weekly basis, and over 80% of our first-time attenders reported that they first attended online. One of the highlights of my ministry there was when we held a baptism where a dozen people were baptized who had first attended the church online livestream (most of them attended for a good 6-9 months online before checking us out in person).”

At the smaller congregation where I currently serve now, our live Sunday morning viewers typically number only 8-10 people. But by week’s end, we see 250-350 total views, and on our most popular topics, as many as 700-800 within a week or two. As Elizabeth Rhyno, our Pastor of Relational Ministries, puts it: “Almost every first-time guest who comes through the doors on a Sunday morning has already watched one or several of our services online.”

The porch, in other words, is full. Many of those people are not ready to come inside yet. But they are there, learning your name, getting a feel for who you are, deciding whether this is a community worth trusting. If there is no front porch to linger on, many of them will simply keep on walking past.

Don’t Drop the Livestream!

Jason Moore and I share a common concern here: there is a growing temptation among weary pastors to drop the livestream now that things feel more “normal.” After years of scrambling to adapt to pandemic-forced changes, of investing in equipment and volunteers and technology they never asked to manage, the impulse to simplify is deeply understandable.

But Moore is clear-eyed about what is actually at stake, and I want you to hear it loud and clear from me. The global pandemic brought online worship to thousands of congregations and into homes for the first time. His argument in Both/And is unambiguous: to build on that momentum, we cannot go backwards.⁶ The worship and ministry of the church going forward cannot be either/or. It must be both/and. And he is equally clear that hybrid ministry is within the grasp of any church willing to pursue it, regardless of size, budget, or technical sophistication.⁶

Dropping the livestream is not a return to normal. It is a decision to close the front porch. Given what we now know about how people discover, explore, and eventually commit to a church community – especially in a post-pandemic culture where online engagement is simply assumed – that is not a pastoral simplification. It is a missional retreat. And that choice will have devastating consequences for the churches that fall prey to it.

What This Means for Discipleship

The shift in attendance patterns raises three honest challenges that ministry leaders need to think through carefully.

1. The first is community and gathering. Online platforms enable genuine connection, but they cannot fully replace the embodied dimensions of physical community. Essential practices like communion, baptism, and fellowship. The spontaneous conversation after the service that turns into a two-hour lunch. The moment someone sitting next to you notices that you seem to be carrying a heavy burden. These are not peripheral aspects of church life. They are part of what it means to be the Body. It’s where those “one another” commands in Scripture are fulfilled. A hybrid ministry mindset honors the digital front porch while actively drawing people toward deeper, in-person roots.

2. The second is pastoral care. It is genuinely challenging to discern someone’s spiritual health or needs when you cannot see them. The screen creates a real perceptual barrier. Leaders who embrace digital ministry need to be actively vigilant about intentionally creating environments where online participants feel safe enough to ask for prayer, admit struggle, and reach out. This often means assigning someone specifically to monitor and respond to comments and prayer requests during services…not as a technical function, but as a pastoral one (in my church, for example, I am one of our livestream hosts, and I engage early in it as part of my pastoral ministry calling). Moore identifies this as one of the most practical and highest-impact investments a church can make in its online congregation.

3. The third is the consumerism question. Not every person who watches your service on Thursday night instead of Sunday morning is treating church like Netflix. Of course some (perhaps many) are. But the anonymity of online participation can easily become a long-term resting place. As I noted in Early Warning: “It is all too easy for parishioners to view worship as a product to consume at their convenience rather than a commitment to participate in with others.” The answer is not to eliminate the convenience. The answer is intentional engagement: we have to find ways to address online participants directly, inviting them into next steps, following up when they do return or occasionally visit, and building pathways that lead from the porch into genuine community.

Rethinking What “Success” Looks Like

Here is the harder conversation that most churches are not yet having: the metrics we use haven’t kept up with the reality we’re experiencing.

For generations, pastoral success was usually measured in a single number – how many people were in the seats on Sunday morning. That number is no longer sufficient. As I insisted in Early Warning, the true measure of impact is not just who is in the pews, but who is being discipled, encouraged, and connected – whether in person or online – and how they are taking next steps in their spiritual and relational growth.

That means learning to track different things. In The Wesleyan Church, my denominational home, we track “peak live viewers.” We even have denominational guidance on this specifically, developed by Matthew Tietje, who currently serves as our Director of Communication and Administration.⁷ That’s total weekly views across all platforms (for example, we stream to both Facebook and YouTube simultaneously, and we have some who attend one and others who attend the other). Omitting our online viewership, or only counting one modality of it, wouldn’t give an accurate picture of our total attendance. Also important are the various aspects of genuine online engagement: comments, prayer requests, follow-up conversations, etc. Digital small group participation. First-time online visitors who eventually show up in person. If you’re not tracking that, you need to…because THAT is where the action is happening. That’s where casual attenders are being formed into disciples (or consumers), and how well we do that is what will shape the church we see in the future.

One more finding from the EPIC research is worth sitting with. Whatever we think of the attendance picture, this is what the data showed among more than 24,000 churchgoers surveyed: more than half reported stronger religious faith (55%) and stronger spirituality (57%) as a result of pandemic experiences.³ The Spirit has been working in the disruption. God was not absent. That is worth measuring, too.

The Mission Hasn’t Changed

The front porch has changed. It’s all digital now. It’s open at midnight now, as well as in the middle of the workday and in the wee hours of the morning. It’s accessible from a hospital bed, a hotel room, or a couch in another state. But the people sitting on it are real, and what they are looking for is legitimate: they’re seeking community, purpose, belonging, and hope.

That is the church’s calling, in every era and every medium, for every generation. The pandemic didn’t create this opportunity. It just reshaped the front porch to enable a space where anyone could find it online easily.

• If you’re ready to think seriously about how to cultivate both your in-person and your online congregation, Jason Moore’s Both/And is the most practical and field-tested resource available (he also does workshops and training on it).

• And if you’re navigating the broader question of how to lead through technological disruption with wisdom and confidence, I wrote Early Warning for exactly that purpose (and I also do workshops and training on that).

The invitation is the same one it has always been. Come on in.

Resources referenced in this article:

1. Tracy F. Munsil, “New Post-Pandemic Research Shows How COVID Rocked America’s Faith,” Cultural Research Center, Arizona Christian University, June 14, 2023, https://www.arizonachristian.edu/2023/06/14/new-post-pandemic-research-shows-how-covid-rocked-americas-faith/.

2. Aaron Earls, “The Post-Pandemic Shift in Evangelical Church Engagement,” Lifeway Research, August 26, 2024, https://research.lifeway.com/2024/08/26/the-post-pandemic-shift-in-evangelical-church-engagement/.

3. Hartford Institute for Religion Research, “HIRR Study Shows How U.S. Churches Have Rebounded Since Pandemic,” Hartford International University, June 17, 2025, https://www.hartfordinternational.edu/news-events/news/hirr-study-shows-how-us-churches-have-rebounded-pandemic.

4. Jason Moore, Both/And: Maximizing Hybrid Worship Experiences for In-Person and Online Engagement (Plano, TX: Invite Press, 2022), Chapter 2. https://amzn.to/4vLf6N8

5. Ann A. Michel, “Reaching People Here and There, Now and Later through Hybrid Worship: An In-Depth Interview with Jason Moore,” Leading Ideas, Lewis Center for Church Leadership, Wesley Theological Seminary, March 14, 2023, https://www.churchleadership.com/leading-ideas/reaching-people-here-and-there-now-and-later-through-hybrid-worship-an-in-depth-interview-with-jason-moore/.

6. Jason Moore, Both/And, Introduction and Chapter 1. https://amzn.to/4vLf6N8

7. Matthew Tietje, “Live Online Attendance Tracking Guidance for The Wesleyan Church,” The Wesleyan Church Resources, https://resources.wesleyan.org/live-online-attendance-tracking-guidance.

Here is what I know: when a new disruptive technology emerges, the optimists always say the same thing.

“Don’t worry. New jobs will replace the old ones. They always do.”

And here is the honest truth: they’re not wrong. But they’re not telling the whole story, either.

This whole line of thinking came into sharp focus for me recently, in a Facebook conversation with my friend and colleague Dr. Ken Schenck, a scholar and self-described futurist. Ken had caught a news snippet making the familiar argument that AI will take a lot of jobs, but will create far more. His response? He wasn’t buying it uncritically.

“I’m a futurist by nature, but what I would really like to know is what these new jobs will be. Because the AI revolution seems to me to be a difference in kind from previous technological advances. I’m very willing to be told what these jobs will be. But I’m just not at all sure that these commentators know what they’re talking about.”

That’s a great question from a thoughtful observer. And as someone who has been following AI development for over 30 years, I had an answer for Ken.

The New Jobs Are Real, and They’re Already Here

The emergence of AI-specific roles is not speculation. It is well-documented and happening right now. Here is a quick cross-section of the new job categories that are rapidly emerging:

• Technical and engineering roles: Machine Learning Engineers, ML Ops Engineers, AI Solutions Architects, Context Engineers, and AI Reliability Engineers.

• Governance, ethics, and risk roles: AI Ethics Officers, Directors of AI Governance and Risk, Trust Engineers, and AI Compliance Officers.

• Strategy and leadership roles: Chief AI Officers (CAIO) and AI Product Managers.

• Human-facing and adoption roles: Human-AI Enablement Partners, Prompt Engineers, AI UX Designers, and AI Content Creators.

• Data and infrastructure roles: Data Stewards, AI Data Curators, Data Center Managers, and AI Solutions Engineers.

• Higher education: Vice Presidents and Chief AI Officers, AI Faculty Fellows, AI Instructional Designers, AI Curriculum Developers, AI Literacy Coordinators, and Academic Integrity and AI Specialists.

I see these listings regularly in my professional networks. I’m even helping mentor several universities through the process of AI integration, and a couple of them have posted such roles. I’ve personally observed a sizeable (300-400%) increase in such postings over just the last two years. The actual aggregate growth is probably closer to 1,000 percent, but I’m only reporting what I can see firsthand.

None of this even touches the scale of traditional roles that are rapidly requiring significant AI fluency: law, medicine, finance, education, and ministry itself. The shift to AI competency as a baseline expectation across virtually every professional field is well underway, and it’s growing.

So yes, Ken, the new jobs are real.

Subscribe now

And Then the Conversation Got More Interesting

After I posted that response, our mutual friend and colleague Miranda Cruz asked a follow-up question that really got me thinking:

“What do you think could be the ratio of jobs created to jobs lost, and non-tech jobs lost to tech jobs created?”

That’s a sharper question. And honestly? It’s the more important one. Not because Ken’s question wasn’t good (it was), but because Miranda’s question forces us to look at the problem from the perspective of the person sitting in the pew on Sunday morning, not just the economist or educator tracking aggregate trends (is it any surprise that she teaches ethics?). But the job loss/gain ratio question isn’t just academic. It’s deeply personal.

My first answer to Miranda was straightforward and honest: The reliable research on this is genuinely complicated.

So I did what any responsible researcher should do: I went to the data. But then I did something else I found equally enlightening. I asked three of the major AI tools the same question to compare their responses, and then I traced their sources (because if you value truth and discernment, the sources always matter more than the AI’s synopsis).

Three AI Tools, Three Very Different Answers

I posed the same research question to Claude (Anthropic), Gemini (Google), and ChatGPT (OpenAI): Is there reliable data on the historical ratio of jobs lost versus jobs created by disruptive technology innovations?

• ChatGPT: Job Losses vs Creation

• Claude: Job losses vs job creation from technological disruption

• Gemini: Technology’s Job Creation Ratio

The results were instructive, and not just for their content. The fact that three AI tools gave meaningfully different answers is itself worth noting. This is exactly why I consistently remind ministry leaders: AI tools are research assistants, not research authorities. The sources they cite matter far more than their summaries.

• Gemini offered the most optimistic framing, presenting historical ratios in a clear comparative table. The Internet and PC era, according to McKinsey’s estimates, yielded roughly a 5.4:1 creation-to-displacement ratio: about 3.5 million U.S. jobs lost, but 19 million created. The automobile era showed approximately 3:1. And the World Economic Forum’s (WEF) 2025 projection for AI and automation comes in at roughly 1.85:1, with 170 million new roles projected against 92 million displaced by 2030.

• ChatGPT was the most cautious, correctly noting that a single reliable historical ratio simply does not exist. It cited the OECD’s finding that across 21 countries, no net job destruction at the national level was observed. But it also surfaced more sobering local data: one additional robot per 1,000 workers in a U.S. local labor market was associated with roughly 5.6 fewer workers employed in that community. The local picture can be sharply negative even when the national aggregate looks encouraging.

• Claude’s analysis was the most nuanced. Drawing on sources including MIT economist David Autor’s longitudinal labor market research and a 2023 peer-reviewed systematic review in Technological Forecasting and Social Change, it surfaced the most sobering finding of all: while the 80-year picture (1940-2018) shows clear net job creation overall, the period from 1980 to the present shows that technology has actually replaced more U.S. jobs than it has generated. And the new jobs are not distributed evenly. Growth is happening at the high-skill and low-skill ends of the labor market, with the middle being hollowed out.

So, is the ratio positive? Historically, yes. Is it reliably positive for everyone, everywhere, on a human timescale? No. And that distinction is really the heart of the matter.

Ken’s Instinct Was Right

Ken’s sense that AI might represent a difference in kind, not just degree, finds real support in the research. And I agree with him.

In the conclusion of my book Early Warning (Invite Press, 2025), I explore a framework developed by Canadian philosopher and media theorist Marshall McLuhan (1911-1980), whose insights remain foundational in media studies and communication theory to this day. His 4 Laws of Media, or “tetrad” of media impacts, asks four questions of any new technology, and they are worth walking through here:

• What does it extend? Every new medium enhances or amplifies some human capacity. A megaphone amplifies a human’s voice; a hammer extends the power of our hand; glasses enhance our vision. We create technology to help us do things we can already do, but better, stronger, or more efficiently. AI extends human cognitive capacity in ways we are only beginning to understand.

• What does it make obsolete? Every new technology replaces another technology and usually makes it obsolete. That doesn’t mean the old technology disappears entirely, but its function changes. The automobile made horse-and-buggy transportation obsolete; they still exist, but primarily for quaint entertainment. AI is making certain cognitive and knowledge tasks obsolete in ways that previous automation waves reserved for physical and routine work.

• What does it reverse into? When pushed to an extreme, every medium will reverse into its opposite intentions. A hammer can build or destroy. Online technology can enable communication or spread manipulation. Social media was designed to connect people, but has also driven isolation and distorted our sense of identity and worth. AI, pushed to its extremes, carries real risks that we are only beginning to grapple with seriously.

• What does it retrieve? Intriguingly, every new technology also retrieves some ancient experience or medium from the past. Email retrieves the telegraph. Movies retrieve campfire storytelling. AI retrieves the scribe, the advisor, the assistant, and the oracle. Those precedents are worth studying carefully.

As I argue in Early Warning, applying McLuhan’s tetrad to AI suggests that its scope and depth of transformation puts it in a category of its own. In terms of paradigm-shifting impact, I would place generative AI alongside only five other innovations in all of recorded human history: the wheel, the industrial revolution, the computer, the Internet, and the smartphone. Every other technological innovation’s impact pales in comparison to the transformative, civilization-reshaping impact those five had. Generative AI deserves spot number six on that list.

What makes this wave genuinely different is something the research confirms directly: unlike previous waves of automation that targeted physical and routine tasks, AI is targeting cognitive and knowledge work. The white-collar expansion that absorbed displaced workers in prior waves of disruption may not buffer the impact this time around. That’s a significant structural change, and it’s why Ken’s concern deserves to be taken seriously.

This Is Where You Come In

Here is the question Ken and Miranda’s exchange is really pointing toward, once you strip away the statistics:

When the jobs disappear, who will be there for the people left behind?

The aggregate data may ultimately be positive. But that is cold comfort for the 55-year-old administrative assistant whose position has been automated, who doesn’t have the resources or time to retrain for an AI engineering role. The time lag between job loss and job creation is measured in years, sometimes in decades. And individuals don’t live inside aggregate data. They live inside their own zip code, their own household budget, their own anxiety about next month’s rent.

In Early Warning, I identify five overarching trends that will shape the future of the Church in the coming years. Trend #5 is directly relevant here, and I want to quote it at some length because I think it deserves to land fully:

“Numerous economic shifts, automation, and changing workforce patterns will influence how people engage with church and ministry. As governmental and corporate entities downsize and restructure, you can expect that many who attend your church will be laid off. As AI use increases, many entry-level roles will be replaced. The more that technology and disruptive innovation reshapes the workforce and job market, the more it’s going to become imperative for churches to not only be aware of and perceptive to that need, but responsive in practical, tangible ways. After all, a church that preaches truth and holds people to high standards is one thing (and a good thing), but if the messaging and services that the church provides don’t provide meaningful and relevant guidance and support through the challenges and changes of life, it isn’t going to last in the long run, and it’s not going to result in transformed lives.”

That’s not alarmism. That’s pastoral realism. The people in your pews are not economic abstractions. Some of them are already feeling this. More will be soon.

And here is the opportunity that rarely gets named in this conversation: the local church, at its best, is precisely the kind of practical biblical community that is positioned to help. Not to solve the societal macroeconomic problem. But to walk alongside the individuals who are caught inside it.

What This Looks Like in Practice

Again drawing from my Conclusion in Early Warning, here are three starting points for ministry leaders who want to lead responsibly in this moment:

1. First, take the concern seriously, and don’t rush to reassure. “New jobs will emerge” is true, but that is not effective pastoral care (instead, it’s a trite platitude with shallow tangible hope). There is a growing need and opportunity for the church to provide tangible community impact through responses to felt needs like job training, financial literacy programs, and helping people navigate AI and workplace change. The transition pain is real, even if the long-term picture is ultimately positive. Meet people in the pain first.

2. Second, know your congregation’s vocational landscape. Do you know which of your members work in fields with high automation exposure? Administrative roles, customer service, data entry, paralegal work, and mid-level accounting are all in the crosshairs. Awareness puts you in a position to walk alongside people before a crisis, not just after. You may not be able to solve the economic problem, but you can make sure nobody faces it alone.

3. Third, consider whether your congregation could become a hub for reskilling. Job transition workshops, digital literacy training, partnerships with local community colleges or online learning platforms: these are not beyond the reach of most churches. In fact, I note in Early Warning that the majority of effective, growing church plants in recent decades did so, at least initially, by meeting one or more specific felt needs in their community that existing churches weren’t meeting. Helping people navigate the AI-driven job market may be one of the most significant felt needs of the next decade. If no one in your community is meeting it, you’ve found an ideal opportunity.

My own job was eliminated due to downsizing this last year, and it happened to my best friend and at least half a dozen others who have experienced similar. The impact is real.

Don’t forget John Wesley’s legacy of innovation here! He directly addresses real life challenges of access to healthcare and medicine in his Primitive Physick (a health and wellness manual). Also, Wesley’s “New Room” in Bristol, which he established to disciple and pastor the fruit from George Whitefield’s ministry and Wesley’s own “field preaching” to the colliers (coal miners) at nearby Hannam Mount, was built to be far more than a worship center. As I explain in my journal article in Didache: “It was a place to help people with their finances, with medicine (herbal remedies), with other life challenges, and help them get back on their feet; the facility included rooms for his lay preachers to get rest and recuperation while in between circuits, resources for study and ministerial preparation, and more.” It is the oldest Methodist meetinghouse in the world, and was literally the cradle of the Methodist movement.

The τέκτων (tektōn) framework I work from is rooted in the image of Jesus as a craftsman who worked skillfully with the tools of his day. That posture, wise workmanship in the service of others, is what ministry leadership in the age of AI needs to model. As I close Early Warning: “The future belongs to those who see disruption not as a threat, but as an opportunity to reimagine how the timeless truths of the Gospel can be shared and lived out in this new era.”

The jobs question is real. The disruption is real. The time lag between loss and recovery is painfully real. And the opportunity for the Church to be a stabilizing, equipping presence in people’s lives during one of the most significant economic transitions in modern history?

That opportunity is very real, too.

Resources referenced in this article:

Acemoglu, Daron, and Pascual Restrepo (June 2020). “Robots and Jobs: Evidence from US Labor Markets.” Journal of Political Economy 128, no. 6: 2188–2244. First published as NBER Working Paper No. 23285, March 2017. https://doi.org/10.1086/705716.

Atkinson, Robert D., and John Wu (May 2017). False Alarmism: Technological Disruption and the U.S. Labor Market, 1850–2015. Washington, DC: Information Technology and Innovation Foundation. https://itif.org/publications/2017/05/08/false-alarmism-technological-disruption-and-us-labor-market-1850-2015/.

Autor, David, Caroline Chin, Anna Salomons, and Bryan Seegmiller August 2024. “New Frontiers: The Origins and Content of New Work, 1940–2018.” The Quarterly Journal of Economics 139, no. 3: 1399–1465. https://doi.org/10.1093/qje/qjae008.

Claude (Anthropic), Gemini (Google), and ChatGPT (OpenAI). Responses to research query on historical job displacement ratios. Research dialogues with the author. March 2026.

Georgieff, Alexandre, and Anna Milanez (January 2021). “What Happened to Jobs at High Risk of Automation?” OECD Social, Employment and Migration Working Papers, No. 255. Paris: OECD Publishing. https://doi.org/10.1787/10bc97f4-en.

Hötte, Kerstin, Melline Somers, and Angelos Theodorakopoulos (September 2023). “Technology and Jobs: A Systematic Literature Review.” Technological Forecasting and Social Change 194: article 122750. https://doi.org/10.1016/j.techfore.2023.122750.

Manyika, James, Susan Lund, Michael Chui, Jacques Bughin, Jonathan Woetzel, Parul Batra, Ryan Ko, and Saurabh Sanghvi (December 2017). Jobs Lost, Jobs Gained: Workforce Transitions in a Time of Automation. San Francisco: McKinsey Global Institute. https://www.mckinsey.com/featured-insights/future-of-work/jobs-lost-jobs-gained-what-the-future-of-work-will-mean-for-jobs-skills-and-wages.

Moretti, Enrico (May 2010). “Local Multipliers.” American Economic Review 100, no. 2: 373–377. https://doi.org/10.1257/aer.100.2.373.

Schenck, Ken. 2026. “I caught a snippet saying that, yes, AI will take a lot of jobs, but it will create a whole lot more.” Facebook, March 12, 2026. https://www.facebook.com/ken.schenck.7/posts/pfbid02c61DLG3dHcFNaJaUnnSv6LJ7XjsJzRMCfRjYdMphdU73wtKWLvDwYxG73RMLk4kml.

Swisher, David J. (2025). Early Warning: How to Lead with Wisdom in Times of Technological Disruption (Invite Press), pp. 169-181 (Conclusion).

Can a machine preach?

It’s not a hypothetical anymore. Generative AI can produce a three-point sermon with Scripture references, illustrations, and a closing prayer in about the time it takes you to find your coffee cup. I’ve seen it. I’ve tested it. The output is often coherent, occasionally even moving.

So I’ll ask again. Can a machine preach?

My friend and mentor Leonard Sweet has a way of cutting through a question like that. In announcing the project I’m about to share with you, he wrote this:

“In an age where machines can generate sermons in seconds but cannot imagine the kingdom for a millisecond, the church needs preachers who can do what algorithms never will: see in the dark and speak what isn’t yet.”

That’s the tension this article is about. And it’s the tension that led me to say “Yes!” to one of the most exciting invitations of my career.

Two Owls

Shortly after Christmas, Len reached out to me. He had read (& endorsed) my book Early Warning and felt it echoed the themes he had been teaching for years at the intersection of imagination, semiotics, and ministry. He wanted to know if I would co-teach a doctoral cohort with him at Northwind Theological Seminary. I said “yes” before he even finished the sentence.

The image at the center of this cohort is Athena, the Greek goddess of wisdom, with two owls perched on her shoulders. In classical mythology, Athena kept an owl on her shoulder because it could see what she could not: the world behind her, the darkness she faced away from. The owl gave her total perception.

For this cohort, we need two such owls.

• The first owl is Sacred Imagination: the capacity to see what isn’t yet, to speak in the Spirit’s native tongue of image and story, to do what Jesus did when He conjured the Kingdom through parable and metaphor.

• The second owl is Responsible AI: not a replacement for the preacher’s soul, but a tool of analytical reach and research depth, used with discernment and clear ethical guardrails.

Together, they give the preacher what Len calls “double vision”: the foresight to imagine God’s future, the insight to navigate our digital present, and the discernment to use tools without becoming one.

A Fair Question

When I shared this with my pastor, he was genuinely enthusiastic about the opportunity, but also cautious - and honest - about the mythological imagery. He told me the Greek imagery made him pause a little. He said it felt like an unexpected approach for a ministry program. I appreciated that reaction, because he’s probably not alone.

So let me address that potential concern directly.

Nobody is asking you to embrace Greek mythology as a theological framework. Athena is a symbol, not a creed; it’s what her owls represent (and why) that matters. The owls are a metaphor. And the Church has a long, honorable tradition of doing exactly what this imagery does: drawing on the wisdom of the surrounding culture to illuminate truth that transcends it.

• Consider the apostle Paul on Mars Hill. In Acts 17, he stood before the Athenian intellectual elite, and said, “Men of Athens, I notice that you are very religious in every way, for as I was walking along I saw your many shrines. And one of your altars had this inscription on it: ‘To an Unknown God.’ This God, whom you worship without knowing, is the one I’m telling you about” (Acts 17:22-23, NLT). And then, in his talk about Jesus, he quotes Cretan philosopher-poet Epimenides “in him we live and move and have our being” and from the Cilician poet Aratus “We are his offspring” (v. 28). Notice that Paul didn’t endorse their philosophy or syncretize their views with his. Instead, he simply used their language as a bridge to communicate a greater truth. He met them where they were and pointed them somewhere they hadn’t yet been.

• This is the same kind of semiotic awareness and communication we see embodied with Patrick, the missionary bishop to the Irish Celts in the 5th century. As I explained in a scholarly journal article, “When Patrick learned that, in Irish mythology, gods and goddesses often manifested in threes, and that one motif even portrayed a tri-faced god, he saw in this an opportunity and a means of communication, rather than an idolatry to oppose.” Also, he added the cross to the pagan sun circle, forming what we know today as the Celtic cross: “The pagan circular symbol Patrick Christianized with a new meaning (wholeness and the beauty of God’s creation),” and to show Christ’s authority over all of nature (Swisher, Evangelizing Post-moderns: A Celtic Model, WTJ 47(2)).

• John Wesley did the same thing in the 1700s. Wesley read voraciously across classical literature, philosophy, and science. He believed that all truth, wherever it was found, belonged to God. He urged early Methodists to develop their minds, engage the culture, and bring every insight into the service of the gospel. His vision of practical holiness was never a retreat from the world. Rather, it was a deeply engaged, wisely equipped presence within it. Wesley scholar & biographer Ken Collins says this was “For the sake of giving the gospel as wide a hearing as possible” (Collins, A Real Christian: The Life of John Wesley, 72).

Unfortunately, we don’t use cultural symbols and visual metaphors often enough in our communication. Even though it’s the language our culture uses and resonates with, our apprehension about being misperceived as secular or worldly often holds us back from genuinely connecting with people in ways they would readily understand.

These two owls stand squarely in that tradition. Imagination and AI are not a theology. They are tools. The preacher holds them both, uses them wisely, and remains irreducibly human at the center.

Subscribe now

Why John Wesley Would Recognize This

Although the imagery comes from ancient Greece, there’s something about the two-owls framework that feels distinctly Wesleyan to me.

After all, John Wesley held together two things that we often pull apart: reason and evidence. He also trusted experience and the Spirit’s movement in the human heart. He didn’t see those as competing claims; rather, he saw them as necessary partners in the pursuit of holiness and truth.

The owl of sacred imagination maps directly onto what Wesley called the affective life (inner emotions), coupled with what he called “works of piety and mercy” (outward actions): the heart set on fire, the Spirit bearing witness – the preacher formed from the inside out. The owl of responsible AI maps to Wesley’s insistence on reason, method, and practical wisdom in service of the mission.

Hold them apart, and you get either sentimentalism or an algorithm. Hold them together, and you get something the church desperately needs right now.

A Vision That Was Ready for This Moment

My colleague and longtime friend Rick Bartlett, who wrote the Foreword for Early Warning (and who studied with Len in an earlier cohort at George Fox University), told me when he heard about this that he was glad to see Len’s vision for this kind of formation finally find its right home. It has been looking for one.

Len has described me in his public announcement of this cohort as “the storm chaser the church needs right now,” and elsewhere as “the church’s current expert on responsible AI.” I certainly wouldn’t claim either of those titles for myself, but I’m honored, and I receive them as a charge, not a credential. They represent a responsibility I take seriously.

The doctoral cohort we are forming together at Northwind Seminary for this Fall is built around exactly that charge. It will include immersive experiences at Orcas Island in the Pacific Northwest, as well as Oxford. Orcas Island is one of the most strikingly beautiful places I’ve been, and it’s a haven for artists and artisans. Len knows that island well; it’s his home, and I know it well from meaningful experiences there in my doctoral program under Len’s mentorship.

Sitting by a fire there with the water and the trees and the night sky around you, you understand something about what it means to see in the dark. Oxford follows in year two. And for your final advance, you get to choose your destination.

This is formation, not just information. It is exactly what the church needs for preachers who will navigate the AI revolution without losing their souls.

An Honest Invitation

We realize this program isn’t for everyone, and that’s OK (in fact, it’s as it should be). This program is for a specific kind of ministry leader: someone who already senses that imagination and discernment are the irreplaceable edge, and who is hungry for spiritual and ministerial formation that takes both seriously.

If any of that is stirring something in you, that stirring is worth paying attention to.

And if it isn’t for you, I ask that you consider who you know. Rick’s first response when he heard about this was:

“Let me know when it’s ready so I can recommend students.”

That’s the spirit! In fact, as you read this, you probably know someone whose call sounds exactly like this. I hope you’ll pass this along to them.

You can read Len’s full announcement on his Substack, and find program details and admissions information at Northwind Theological Seminary’s page for the cohort. Both links are below.

Machines can write sermons. Only humans can preach. Let’s make sure the Church knows the difference.

Resources referenced in this article:

• Acts 17:16-33 (NLT).

• Collins, Kenneth J. (1999), A Real Christian: The Life of John Wesley, (Nashville, TN: Abingdon), 72.

• Northwind Theological Seminary, “Semiotics: Sweet Fellows Doctoral Cohort in Homiletics,” northwindseminary.org, 2026. https://www.northwindseminary.org/semiotics-2026

• Swisher, David J. (2012). "Evangelizing Post-moderns: A Celtic Model." Wesleyan Theological Journal, 47 (2) Fall 2012, 179-194. https://tinyurl.com/WTSPatrick

• Leonard Sweet, “Preaching with Double Vision in an AI World,” Substack, February 2026.

Leonard Sweet

Preaching with “Double Vision” in an AI World

The “information-only,” pulpit-centric sermon is over. If a machine can scrape the internet, find three points, and generate a sermon in five seconds, why should a congregation listen to a human do the same…

Read more

4 months ago · 8 likes · 5 comments · Leonard Sweet

In Part 1 (last week), we explored Paul Hiebert’s centered-set framework and why it resonates so deeply with Wesleyan values. The core insight is this: instead of defining ourselves by rigid boundaries (what’s “in” or “out”), centered-set communities orient everything around a center, Jesus Christ, and ask which direction people are moving.

It’s a powerful lens for thinking about church membership and discipleship, and those applications are well-documented not only by Hiebert’s own writings, but especially for contemporary culture by Daniel Im and Mark Baker.

But I want to suggest something more. This framework might be the most practically useful tool you have for navigating one of the most pressing challenges ministry leaders face right now: the question of how to engage responsibly with emerging technologies, particularly rapidly-evolving ones with urgency like artificial intelligence.

The Failure of Binary Thinking

Unfortunately, most religious conversations about technology tend to default to bounded-set thinking without realizing it. If you’ve been in a board meeting or commented on Facebook where newer technologies and their use in the church were discussed, you’ve likely seen this firsthand:

• AI will either usher in unprecedented human flourishing, or it will spell the end of humanity as we know it.

• Social media is either a powerful tool for gospel proclamation or the root of all relational breakdown in our culture.

• That new platform is either a gift from God or a digital threat to be resisted.

This kind of binary thinking serves no one well because it prevents nuanced engagement. It forces false choices (often utilizing logical fallacies). And it leaves ministry leaders paralyzed, either resisting every technological change out of fear (bounded-set rigidity) or uncritically adopting every new tool without stopping to reflect (fuzzy-set tolerance).

Neither posture honors the complexity of the moment we’re in.

Subscribe now

The Better Questions

A centered-set approach to technology doesn’t ask “Is this tool inside or outside the boundary of acceptable Christian use?” That forces a false binary that ascribes moral intentionality to inanimate objects.

Instead, a centered-set approach asks: “Does our use of this technology move us toward Christ, or away from him? Does it deepen our capacity for love, justice, and mercy...or diminish it? Does it enable us to serve people well or better, or merely exploit them?”

Those are different questions. And they open up a much more useful conversation.

Here is what this looks like in practice, across five themes that deserve ongoing attention from every ministry leader who is actively engaging with emerging disruptive technology like metaverse & VR, cryptocurrency, or generative AI:

1. First, centered-set thinking resists what I would call technological determinism. The bounded-set version of AI anxiety tends toward fixed predictions: either AI will inevitably solve all human problems, or it will inevitably destroy us. Centered-set thinking rejects both extremes. The future is not fixed. Human agency matters. We can make choices that move us toward more Christ-like uses of AI, tools that serve human flourishing, enhance justice, protect dignity, and extend mercy.

2. Second, if Christ is our center, then human dignity, grounded in the imago Dei, becomes non-negotiable. Every AI application, every algorithm, every automated decision system must be evaluated through this lens. Does it honor human dignity or erode it? Does it enhance human agency or diminish it? Does it serve the vulnerable or exploit them? These questions create clear direction without demanding premature answers to questions that are still unfolding.

3. Third, centered-set thinking creates space for learning. The rapid pace of AI development means we cannot wait until everything is figured out before we engage. Instead, we must learn by doing. A centered-set approach gives permission for experimentation, iteration, and honest course correction. We try new things. We evaluate whether they’re moving us toward Christ-like ends. We adjust. This is not recklessness; rather, it’s faithful stewardship in an uncertain moment…just like the disciples and the early church leaders had to do in the years following Christ’s death and resurrection.

4. Fourth, that openness must be grounded in theological depth. You cannot orient toward a center unless that center is clearly defined. Navigating AI responsibly requires robust theological convictions: a clear anthropology (who are humans before God?), a clear eschatology (what is our ultimate hope?), and a clear missiology (what is the church’s purpose in the world?). These convictions are the ballast that keeps us stable when the technological waters get rough.

5. Fifth, theological optimism about grace and human potential should fuel healthy innovation, while theological realism about sin provides necessary caution. We believe God is doing new things. We believe human creativity, rightly directed, can serve God’s purposes. But we also know that technologies amplify both our capacities for good and our capacities for harm. Innovation requires wisdom. The centered-set approach holds both in tension: openness to what’s new, grounded in what’s ancient.

What This Means for the “Nones” and “Dones”

Research consistently shows that the fastest-growing religious category in North America is the “nones,” those who claim no religious affiliation. Many of them are also “dones,” people who were once active in churches but left, often wounded by bounded-set rigidity or bored by fuzzy-set meaninglessness.

If you hear people talk about “deconstructing” their faith, or use words like “ex-Evangelical” or “neo-Christian,” that’s the latter group. They got hurt in the crossfires of bounded-set approaches, or became disillusioned by the theologically light weaknesses of fuzzy-set approaches.

But a rapidly-growing segment of American culture has NO religious affiliation. In former years, a large number of people would answer censuses and surveys saying they were “Baptist” or “Catholic” or “Methodist,” even if they rarely ever attended. Their parents attended the church, or they were raised in it, or they perhaps attended at Christmas and Easter, but that was about the extent of their involvement. That kind of cultural Christianity played a significant role in shaping American Christianity throughout the 70s, 80s, and 90s, and that has led us to the place where we are now in the 2000s, 2010s…and especially so in the 2020s.

But today, it is far more common to have a huge (& growing) segment of the population that has never darkened the door of a church except maybe for a wedding or a funeral. They weren’t raised in Sunday School, didn’t go to VBS, and didn’t have a church-going family. Church was just never a thing for them. Their social calendars had plenty of other sports and activities, but no church events, and no reason to include them. In the rapidly-growing suburban community where I live, there are over 30,000 people who identified as “nones” within a 5-mile radius of my church. On censuses and survey, they claimed NO religious affiliation whatsoever.

Centered-set communities offer something genuine to both groups.

To the “nones” who have never experienced authentic Christian community, the centered-set approach says: “Come, belong with us.” You don’t need to have everything figured out. You don’t need to believe what we believe yet. Just join us on this journey toward Christ.”

Side note: If you cringed when you saw my mention there of belonging before believing, that’s a sure sign that you’ve been operating under bounded-set thinking, which inevitably leads to “in or out” logic (or worse, “us vs them” communication).

To the “dones” who fled bounded-set legalism, centered-set communities offer freedom and grace without abandoning truth. To those who left fuzzy-set churches that lacked any compelling purpose, it offers clear direction and meaningful transformation.

And here’s where technology becomes a genuine ministry asset rather than just a logistical tool. Digital platforms, when used with centered-set intentionality, can serve as extraordinary points of first contact and genuine belonging. Not as a replacement for embodied community, but as a front porch. A place where people who are moving toward Christ, even if they can’t yet name what they’re moving toward, find a welcome.

I write about this extensively in Early Warning, particularly in my chapter on livestreaming and online platforms, which today are the front porch of the church, and which the vast majority of people check out first before ever showing for a worship service.

The One Question That Matters

Here’s the thing about centered-set thinking. It doesn’t give you a list of approved and banned technologies. It doesn’t tell you which AI tools are safe and which ones to avoid. I know that can be frustrating for people who want hard boundaries, but that’s simply not what it’s for…nor is it how technology (or discernment) works.

What it gives you instead is a posture. And a question.

The question is this: Are we moving toward Christ, or away from Him? And closely related to that is, “Does this technology enable me (or the people we serve) to move toward Christ, or lead us further away from Him?

For individuals, for communities, for our engagement with technology, for our mission in the world, that is the centered-set question. It applies to your sermon prep workflow. It applies to your church’s social media strategy. It applies to how you think about AI-generated content, automated giving platforms, acceptance of cryptocurrency, livestreaming quality, and every other technology decision you’ll face in the years ahead.

You may not have a clear answer every time. And often it’s how you use it that determines the answer. But asking the right question is always the right place to start.

Resources referenced in this article:

• Paul G. Hiebert, ‘The Category Christian in the Mission Task,’ International Review of Mission 72, no. 287 (1983): 421-427.

• Mark Baker, Centered-Set Church: Discipleship and Community Without Judgmentalism (IVP Academic, 2022).

• Ryan P. Burge, The Nones, Second Edition: Where They Came From, Who They Are, and Where They Are Going (Fortress Press, 2023).

• David Swisher, Early Warning: How to Lead with Wisdom in Times of Technological Disruption (Invite Resources, 2025).

Here is a scenario that might feel a bit familiar.

A new technology tool shows up on your radar. Perhaps a colleague sends you a link to an AI tool for sermon prep. Maybe your worship team starts talking about a new platform. You find out about an amazing tool or app that seems “too good to be true,” or which does more than you ever thought possible. Or someone in your congregation or leadership team asks whether your church has a policy on a particular app.

And almost immediately, a question forms in your mind.

“Is this okay... or not?”

That instinct to sort things into “approved” and “forbidden” categories is deeply human. It’s also natural: we are trying to protect our congregations. We want to be wise stewards. And we’re working fast, because technology doesn’t wait for us to catch up.

But after decades of working at the intersection of ministry and technology, innovating regularly with technology in higher education, then waiting on ministry contexts to finally catch up, I am convinced that the “in or out” question is the wrong question to ask. Unfortunately, the framework most of us tend to default to when evaluating technology is quietly working against us, usually without us even realizing it.

There is a better map. It comes from a missiologist that many pastors and ministry leaders may have never encountered, but one whose influence and thinking I deeply admire and utilize often. And once you see it, you’ll probably start applying it everywhere.

Subscribe now

Paul Hiebert and the Centered-Set Model

Paul Hiebert was a missionary anthropologist who spent his career thinking about how communities define themselves and welcome others. His research led him to identify three fundamentally different ways groups organize their membership, which he called “bounded sets,” “fuzzy sets,” and “centered sets.”

He developed this framework to help missionaries think about conversion and church membership. But it turns out to be one of the most practically useful lenses I’ve found for thinking about technology, innovation, and ministry leadership.

Here’s a quick tour of all three...

1. Bounded-set communities define themselves by clear lines. You’re either inside or outside. Membership depends on conforming to specific criteria: the right beliefs, the right behaviors, the right standards. In bounded-set churches, the sequence is typically, “Believe, then Behave, then Belong” (in other words, before you’re welcomed into community, you need to get your doctrine right and your life in order). Bounded-set thinking provides clarity, and that’s important. But it also tends toward legalism, fundamentalism, and judgment. It creates an inevitable “us vs. them” mindset because if you’re “in,” then you’re “out” (and therefore not one of us). Unfortunately, it frequently confuses cultural preferences with biblical essentials, making it very difficult to tell what truly matters from what is simply tradition. The history of Western evangelicalism is littered with the aftermath of this approach, and much of what we call “deconstruction” today is the fallout from those who suffered through this approach in their church upbringing. It is neither missional nor redemptive.

2. Fuzzy-set communities react against that rigidity by eliminating meaningful distinctions altogether. Boundaries go soft. Tolerance becomes the supreme virtue. Theologian Mark Baker has observed that the result is often a vague Christianity that isn’t particularly interesting, life-giving, or transformative. Without clear direction, people drift. Fuzzy-set thinking avoids judgment, but it also avoids guidance. It all-too-easily becomes an “anything goes” approach that is often lukewarm and full of watered down Christianity with unqualified acceptance. It’s essentially, “You’re in because you claim to be one of us (whether you ever have any desire to live or be like us or not).”

3. Centered-set communities offer a third way entirely. Rather than defining a boundary, they define a center, in our case, Jesus Christ. What matters is not where someone is standing at any given moment, but in which direction they’re moving. Someone who is far from Christ but genuinely moving toward him is welcomed and affirmed. Someone who appears close but is drifting away needs gentle redirection.

The sequence for this approach typically becomes “Belong, Believe, Become.” People are welcomed into the community first; they find acceptance, joy, love, connection, etc. And then in that context of authentic belonging, they encounter Jesus at some point and begin to believe. Over time, as they keep moving toward Christ, transformation happens.

Hiebert described this as “hard center, soft edges,” which stands in sharp contrast to bounded-set thinking’s “hard edges, soft center.” The theological convictions about Christ remain non-negotiable. But the boundary around who can explore faith stays open.

A great synthesis and application of Hiebert’s “centered set” framework for church contexts is Daniel Im’s No Silver Bullets: Five Small Shifts that will Transform Your Ministry (B&H Books, 2017), which I thoroughly enjoyed, and the more recent Mark Baker’s Centered-Set Church: Discipleship and Community Without Judgmentalism (IVP Academic, 2022).

Why Wesleyan Communities Already Know This

I want to be clear about something. Hiebert’s framework emerged from Anabaptist missiology, not Wesleyan theology. And admittedly, these are distinct traditions with different histories, although we do share many commonalities.

But here’s what I keep coming back to: when I first encountered Hiebert’s centered-set model, it felt less like a new idea and more like a description of what Wesleyan communities look like when we’re being most faithful to our own heritage.

For Wesleyans, it starts with prevenient grace – the grace that goes before. John Wesley’s conviction was that God is already at work in every human heart before any conscious movement toward faith. The Holy Spirit precedes our response, drawing and enabling. If you want a Baptist version of this, think of Henry Blackaby’s Experiencing God approach. This maps almost perfectly onto centered-set thinking, which recognizes that people can be moving toward Christ even before they fully understand that’s what is happening. And the best part is that we don’t create that movement, and it doesn’t depend on us. Instead, we discern and cooperate with what God is already doing.

Then consider Wesley’s understanding of sanctification as not just an event, but an ongoing journey. “Going on to perfection” is not a destination you arrive at and check off. It is a direction you commit to for a lifetime. Eugene Peterson’s phrase echoes this thinking well: “a long obedience in the same direction.” Many concede that you will likely never fully arrive this side of glory, but you keep moving toward Christ because the destination is what matters (along with the trajectory). That is, at its heart, the centered-set vision.

And then there is Wesley’s insistence that “there is no holiness but social holiness.” He refused to separate personal transformation from social engagement. Centered-set thinking holds the same balance: it resists the bounded-set temptation toward private piety disconnected from public concern, and it resists the fuzzy-set tendency toward social activism disconnected from the gospel’s transforming power.

This is not an Anabaptist idea mapped awkwardly onto a Wesleyan framework. It is a language for something Wesleyan communities have always practiced at our best.

A Shift Already in Motion

If you’re part of a Wesleyan context, you may have noticed that The Wesleyan Church restructured its membership and discipleship pathways many years ago, moving deliberately to “Belong, Believe, Become” in our membership pathways and materials.

That’s not a minor administrative update. It’s a theological conviction finding institutional form and practical reality. It reflects what research consistently demonstrates: that people today don’t first decide what they believe and then go find a community that matches. They find a community where they experience genuine love and encounter lives transformed by Christ, and then they begin to explore faith there once it feels safe and welcoming. Transformation follows relationship.

That is the centered-set pattern.

And it is exactly the posture we need as we engage with technology and innovation in ministry. For example, I am far less concerned that an app, tool, or product was designed and marketed by Christians, or that it lines up with every theological foundation and believe I have. I don’t get to impose my rules on third-party products and tool. However, I do care – very much – what principles it operates by, and what its trajectory is like. As we’ve seen in multiple cases in recent weeks (one of which I wrote about in my inaugural post), a company’s ethics and values may not be that significant in the day-to-day operation of products and functions, but they absolutely DO matter when push comes to shove…when integrity is on the line.

Coming in Part 2

So what does any of this have to do with your church’s AI policy, your social media strategy, or your decision about whether to experiment with a new platform?

More than you might think.

In Part 2 (next week), I’ll show you how centered-set thinking reframes the technology question entirely. Instead of asking “Is this tool safe or dangerous?,” you’ll find yourself asking far more useful questions that will actually help you lead well. I will also explore what this framework means for reaching the growing number of “nones” and “dones” in your community, as well as for the challenge of forming disciples in a digital age.

The centered-set approach doesn’t make the hard decisions disappear. But it gives you a much better map for navigating them, because bounded-set rules can change in a heartbeat as companies and their cultures and leadership shift. But when you prioritize destination logic over inclusion/exclusion logic, you get useful ways to solve challenges that don’t compromise your mission and values.

.

Resources referenced in this article:

• Paul G. Hiebert, ‘The Category Christian in the Mission Task,’ International Review of Mission 72, no. 287 (1983): 421-427.

• Mark Baker, Centered-Set Church: Discipleship and Community Without Judgmentalism (IVP Academic, 2022).

• Eugene Peterson, A Long Obedience in the Same Direction (Downers Grove: InterVarsity Press, 1980).

• Daniel Im, No Silver Bullets: Five Small Shifts that will Transform Your Ministry (B&H Books, 2017).

Picture this.

You are sitting in a meeting. A colleague across the table is sharing something important...maybe a pastoral concern, or perhaps a ministry idea they have been developing for months. Or a congregation member approaches you to greet you, then shares a concern with you. And then it happens. A phone lights up. A laptop chimes. Someone’s eyes drift down, fingers begin to move, and just like that, the moment is gone.

I have been in the room when it happened more times than I care to admit. I have watched the person on the receiving end of this technology distraction notice it, too. There’s usually a fleeting, almost imperceptible flicker across their face…the slight recalibration of what they were willing to say next. They understand, but they became more hesitant to share now with the awareness of divided focus. I have also watched the dynamic and tenor of an entire meeting change because of it.

Personally, I love laptops, and I use one regularly. I prefer a larger screen and a full keyboard when I’m working. However, when I am trying to meet with someone, such devices are counterproductive. In meetings, even when the person opening one genuinely intends to take notes, it creates a wall. Literally. It erects a physical barrier between you and the people around you. Eye contact disappears. Body language begins to seem closed off. And everyone else in the room wonders, reasonably, whether you are actually paying attention or just catching up on email or scrolling social media. The perception problem is real, even when the intent is good.

Subscribe now

Please understand: I am not making a case against technology in meetings (I am all for it). Rather, I am making a case for choosing the right tool for the work you are actually doing. And for ministry leaders especially, the work of a meeting is almost always relational before it is ever administrative.

That distinction changed how I thought about what I carry into a room.

The way my mind works

For years, I have searched for a digital note-taking solution that actually worked for the way my mind works.

Here is what might be helpful to know about me: I have a unique subset of eidetic memory that is specific to things I write, sketch, or draw by hand. I like to call it videographic, but in short, it means that whatever I write down, I remember and can recall the precise details of. So, I can peruse a stack of my own handwritten notes and quickly locate information in seconds (without reading any of it), not because I remember the words, but because I remember the visual. I can remember exactly how it looked, and I can even tell you where on the page my specific note appeared, what color I used, and how I drew it. Years later, I can close my eyes and describe notes I took from a specific conversation or conference session with remarkable accuracy.

That is a unique gift. But it comes with a catch: the moment those handwritten notes get converted into digital text, my memory of them evaporates.

It’s as if the conversion to digital severs the videographic link in my memory. Once my handwritten notes look like every other typed document, I lose that visual anchor. So every app that has tried to “help” me by converting my handwriting to clean, searchable text was counterproductive; it actually hindered the very thing that would have made my notes useful to me.

Another popular technology enablement these days is iPads, tablets, and other larger mobile screen devices, often where you use your fingers to write. I can see the attraction for some, and I’ve tried them myself. However, writing with your finger or a mouse (or even a stylus) on a smooth glass screen feels nothing like writing on paper. The resistance is wrong. The subtle sound is wrong. The whole tactile experience is so artificial that it disrupts the flow of thinking that natural writing creates. I’ve tried EverNote, Apple Notes, OneNote, Penultimate, and more. None of them won me over. None of them came close to replicating the experience of a pen on paper.

And so, for years, I kept carrying a legal pad. That worked...until it didn’t. Eventually, I had a stack of notebooks with no search function and no backup, with no easy way to share or present from them, either.

What changed

A couple of years ago, though, I came across the reMarkable 2 digital paper tablet. I was skeptical. I had been disappointed too many times before. So I scrutinized the fine print, watched every video available, read dozens of reviews. It sounded too good to be true. Fortunately, it wasn’t, and I was not disappointed this time.

What the folks at reMarkable have figured out is that the writing experience itself is the product. They engineered a touch-resistive display with just the right amount of friction, and a stylus with just the right amount of weight and tip feel, so that writing on the device feels and sounds remarkably like writing on paper. Not approximately like paper. It feels like I’m writing on actual paper. The difference from a standard tablet to this is immediate and unmistakable.

And most importantly for me...it preserves my handwriting. My notes look the way I wrote them, and the visual memory stays intact. Even better…it lets me re-size my notes, edit and erase them, and more. I couldn’t do any of that on a notepad with an inkpen.

But it does more than just feel right. I am also able – finally – to organize my notes into notebooks and pages, search through them, and adjust pen sizes and styles (pencil, pen, felt tip, marker, highlighter, paint tip, etc.). I can even group and move sections of handwritten text, and revert through earlier versions if I make a mess of something. If the speaker gives 5 points, and I only left room for 4, I can activate that section of notes, shrink it down just a bit, and add in that 5^th point.

I can also import PDFs for annotation and markup. I can save notes to a shared drive. I can even present from it directly in a Zoom call.

And then there is the stylus itself. I chose the upgraded version with an eraser built into the back end, so I can flip it over and erase naturally, exactly the way you would with a pencil. It is a small thing, but it matters. It is the kind of design detail that tells you the people who built this actually thought about how people write.

Lastly, it has layers. I can pull up a grid, or notebook paper (select the rule width), or blank paper, sketch out ideas, and then…add a new layer, so that colleagues, or my friends, can add their thoughts & ideas, too, with each one’s contributions on separate layers. My daughter loves this feature when we’re dreaming up new cake designs or planning activities: “Here’s what I’m thinking,” I say. “Oh, OK, well here’s what I’m thinking,” she says. And then we can compare and edit them!

A ministry case for single-purpose tools

Now, I know this writeup is a bit of a departure from my norm, as I tend to camp out in themes, implications, and practice aspects rather than products. However, I do want to make a case here that goes beyond mere technology and productivity.

John Wesley was one of the most prolific letter-writers and journal-keepers in the history of Christian ministry. He wrote thousands of letters. He kept detailed journals across decades (we have multiple volumes of them preserved). He understood that the discipline of writing things down was not just a practical habit. It was a spiritual practice. It was how he processed what he was learning, tracked what God was doing, and maintained accountability in his own spiritual growth as well as in his thinking as a minister and leader of a movement.

Wesley was also deeply committed to being present with the people in front of him. The whole Methodist movement was built on face-to-face accountability – the class meeting, the bands, and the society – all of which involved direct and attentive engagement with people at their point of need. He did not let the administrative machinery of a growing movement displace his attention to individuals.

There is a straight line from that instinct to the question of what we carry into a room.

In a culture that rewards multitasking and equates busyness with effectiveness, choosing a single-purpose tool is a quiet act of resistance. When I walk into a meeting with my remarkable digital paper, I bring one thing that does one thing. It does not buzz. It does not notify me. There is no red icon telling me how many missed messages or notifications I have. It does not tempt me to check anything (and when I’m in the creative moment, I really don’t care to see those, because they distract me from my creative thinking process).

It is a note-taking device, and that is all. And because it looks like a slim notebook rather than a laptop, it does not create that perceptual wall, either. That way, people can see that I am present. People can see that I am writing; I’m taking notes on what they’re saying. And that sends a powerful message…that who they are, what they say, and my time with them matters.

There have been so many times I’ve taken it to a pastor’s conference or scholarly gathering and, when I got it out to take notes, it wakes up instantly. No fiddling with cords, bags, screens, or tools.

I was engaged in the moment and attentive, and others could tell. It has been such a powerful counter to what I used to have to do (of pulling a laptop out of a bag, unfolding the screen, and finding space – and often an outlet) – to get to work. And with a laptop, I’m typing, whereas with my ReMarkable, I’m writing…and engaging and participating with others (instead of lost in my laptop’s digital world).

Yet, it still has all of the capabilities of digital text. For those who prefer it, you absolutely can convert your written notes to digital text. You can scan and search, too. There are a dozen or more cool things I can do with the text that I simply can’t do with handwritten notes on a notepad.

But I also think there is a stewardship dimension worth naming. Your notes from a conference, a denominational gathering, a staff retreat, or a presbytery meeting are a ministry asset. The insights you capture there do not belong only to that moment. They inform how you lead, how you preach, how you advise colleagues, and how you make decisions for months afterward. A legal pad that disappears into a filing cabinet is not serving that purpose. A searchable, organized, cloud-synced digital notebook that you can pull up on your phone or share with a colleague...that represents stewardship of ideas.

And for those of us who find that the act of handwriting is itself part of how we think and pray and process...the reMarkable does not ask us to give that up. It just makes sure we do not lose what we write.

One thing well

Now, lest you think this is nothing more than a sales pitch, I want to be honest about what the reMarkable 2 is not.

It is not a tablet. It does not run apps. It does not have email or calendar or a browser. If you are looking for a multifunction device that does everything, this is definitely not it, and if that’s important to you, it will only frustrate you.

But for me...that simplicity is one of its greatest features.

When I go to a meeting, I bring my reMarkable, and I am there. Yes, I’ll likely have my phone, too, but I usually flip it over or put it out of sight. When I am in creativity or planning mode, I simply open my reMarkable, and there is nothing to distract me except ideas and interactions we’re having in the moment. No notifications. No inbox. No pull toward something else. Just the work of creative thinking, and the page, and the pen.

A tektōn chooses tools that fit the work best. For me, my reMarkable 2 fits the work of ministry in ways that a dozen other devices never did for me. It is by far the best technology investment I have made in the last decade, and it has been a gamechanger for how I take notes and engage. I would recommend it without reservation to any ministry leader who has ever felt the tension between staying connected and staying present. You know which one matters more.

Resources

• Overview video: reMarkable 2 on YouTube

• ReMarkable 2 on Amazon

• ReMarkable’s website (for more info)

Note: The article and links above are my personal recommendations because it’s a product I believe in and use often. I have seen firsthand the benefits of this technology. I do have affiliate links enabled, but none of this is sponsored content.

Picture this. A researcher on Meta’s AI Safety and Responsibility team (you read that right, a professional whose full-time job is AI safety) decided to set up a popular new AI agent to help manage her email inbox. What happened next, she described publicly on X: the agent began deleting emails without her approval. She tried to stop it. She couldn’t get it to stop from her phone. In her own words: “I had to RUN to my Mac mini like I was defusing a bomb.”

That story says more about this moment in AI than a hundred explainer articles.

The last week or two has been one of those inflection points that will look obvious in hindsight. Two major developments broke into the broader public conversation almost simultaneously, and both of them are signaling something very mportant for everyone who leads in ministry. The underlying thread connecting them is a single shift that is now accelerating faster than most people realize: AI has stopped merely responding to us, and now it is capable of acting on our behalf. This changes things dramatically for ministry leaders, whether you have been paying close attention to AI or not.

So let me walk you through what happened, what it means, and what a wise response looks like from where we stand.

First, a Quick Orientation: What Is an AI Agent?

Most of us have some experience now with AI tools that answer questions, generate text, or help us think through problems. You type something in, and the AI responds. That chatbot model is what most people are familiar with.

An AI agent, though, is something different. Instead of responding and waiting, an agent can plan a sequence of steps, take action, use tools, and keep going until a task is done, often without you being involved in every step (usually you design it to do specific things for you). It can read your email, browse the web, fill out forms, manage files, schedule appointments, and submit completed work...all on its own.

The key distinction is this: a chatbot tells you things. An AI agent does things. And “doing things” means it needs access to real systems, real data, and real accounts. Which brings us to this week’s news.

Development #1: OpenAI’s GPT-5.4 Thinking Model Brings Agentic AI Mainstream

A few days ago, on March 5, 2026, OpenAI released GPT-5.4, and the headline feature is significant. It is the first “mainline” model from OpenAI with built-in computer use capabilities, meaning it can interact directly with software to complete, verify, and fix tasks in what the company calls a “build-run-verify-fix loop.” It can read screenshots, click buttons, type inputs, and navigate interfaces the way a human would. It can handle a 1-million-token context window, enabling it to manage large, multi-step projects from start to finish.

In ChatGPT, the new GPT-5.4 Thinking version also shows you its reasoning plan upfront, so you can course-correct while it is working, rather than waiting until it is finished and then starting over.

Here is why this matters for ministry leaders specifically. Until very recently, setting up an AI agent typically required using dedicated workflow tools like Make or Zapier, platforms that take real technical skill to configure. GPT-5.4 changes that significantly. The agentic capability is now built into the same interface that you are probably already using. That lowers the barrier to entry considerably. But, as we are about to see, lowering the barrier to entry is a two-edged sword.

Development #2: Einstein, OpenClaw, and the Rise of “Nuisance Tech”

A few days before the GPT-5.4 release, a different story was making waves in education circles. A 22-year-old entrepreneur named Advait Paliwal launched a product called Einstein, built by a startup called Companion. Its marketing was blunt: Einstein would log into a student’s Canvas account every day, watch lectures, read essays, write papers, participate in discussions, and submit homework automatically. “Set him up and forget about it,” the site promised. “Einstein checks for new assignments and knocks them out before the deadline.”

The reaction in higher education circles was immediate and visceral. Educators flooded social media with alarm. IT departments scrambled. My colleagues in higher education had a chaotic week dealing with all of the fallout. The story went viral.

But here is where it gets more nuanced, and also a bit more instructive. My higher education colleague Marc Watkins, a writing and rhetoric instructor whose Substack Rhetorica explores the intersection of AI and education, offered a more measured read of the situation. He argues that Einstein itself was likely more marketing hype than genuine threat, pointing out that it appeared to be a hastily assembled wrapper built on top of an already-existing open-source tool called OpenClaw, and that the site was eventually taken down, not because Canvas or any institution stopped it, but because Paliwal had apparently never checked whether he could trademark the name “Einstein.” Marc’s framing of this as “nuisance tech” is genuinely useful: AI-powered tools that create alarm and disruption not because they represent a fundamental breakthrough, but because they expose how unprepared our existing systems are for what is already possible.

My colleague Michelle Kassorla, who is a faculty mentor (like me) with AAC&U’s Institute on AI, Pedagogy, & the Curriculum, actually tested Einstein before it was taken down. She discovered firsthand that it was essentially a front end for OpenClaw. Her research, along with Marc’s analysis, is excellent reading. If your ministry world involves higher education contexts, I definitely commend their analyses to you.

But here is the real takeaway, and it is not about the Canvas LMS or any specific agent. The issue is this: If a single developer, working alone, can build a tool capable of completing an entire online course on a student’s behalf...then the underlying architecture that makes that possible is already here. Einstein may have been mostly noise, and it has already been taken down, barely a week after it’s release. However, the capability it demonstrated – and the vulnerability it exposed – is very real.

Subscribe now

The Security Dimension: When Agents Go Wrong

OpenClaw, the tool underneath Einstein, is itself worth understanding. It is an open-source AI agent that can browse the web, manage email, control desktop applications, send messages on messaging platforms like WhatsApp and Telegram, and remember interactions across weeks or months. Its creator, Austrian developer Peter Steinberger, describes it simply: “the AI that actually does things” (and thus the title of this article).

That description is both the appeal and the problem.

Cisco’s AI security research team put it plainly: “Security for OpenClaw is an option, but it is not built in.” The product’s own documentation concedes there is “no perfectly secure setup.” Security researchers have documented cases of credentials being exposed, malicious add-ons being distributed through its skill repository, and agents taking actions users never intended or authorized.

Which brings us back to that Meta researcher running to her computer to stop her self-initiated agent from deleting her inbox. Hers is not a cautionary tale about carelessness. After all, she is a professional who studies this for a living, publicly demonstrating how quickly agentic tools can exceed the boundaries their users thought they had set.

There is also a broader dimension that has received less attention. AI agents built on OpenClaw and similar platforms are now active across public social media. In fact, researcher Ethan Mollick, an Wharton School professor whose research I follow regularly, noted recently that he no longer reads the replies to his posts on X because they are overwhelmed by AI-generated responses…not from humans, but from autonomous agents that are allowed to roam freely online.

We already know that much of the discourse, outrage, and disinformation campaigns we have observed in recent political history has been orchestrated (and outright manipulated) by bot accounts – AI agents. So, we may very well be approaching a moment in technology history where the open public web is largely inhabited by bots talking to bots, while genuine human conversation migrates to closed, invite-only spaces.

What This Means for Ministry

I want to be clear about something. Although the Einstein headlines were framed as an education story, it is far more than that, and it has direct application to ministry leadership. The same dynamics apply to virtually any system that manages trust, access, and human interaction. The Canvas LMS simply happened to be the first target. But the platforms ministry uses, church management systems, communication tools, small group apps, online giving platforms, email and calendar systems, are not far behind. For example: Planning Center, Breeze, Church Community Builder, etc. – all can be set up with AI agents, and could be at risk.

Consider a few scenarios that are closer to your world than you might think:

First, think about pastoral communication. AI agents with access to a pastor’s email and calendar are not hypothetical. Microsoft 365 and Google Workspace are already rolling out agentic features. In a context where communication often involves sensitive pastoral matters, confidential prayer requests, and vulnerable people... who has thought through the policies for what an AI agent is and is not authorized to do on a pastor’s behalf?

Second, consider your online community. If Clawbots are already flooding Ethan Mollick’s replies on X, they are almost certainly active in comments sections, group pages, and forums where your congregation engages online. How do you maintain authentic pastoral presence and genuine human community in a space increasingly populated by autonomous agents?

Third, and perhaps most importantly for those of you who lead training programs or ministry formation contexts, the Einstein story is a preview of what is coming. Agentic tools that can complete assigned tasks without active human engagement will not stay in Canvas. Professional development and certifications? Yes. Ministry credentialing attempts? Likely. They will find every online environment where credentialing and completion matter more than genuine formation. That is a question worth considering.

The Discernment We Need

Here is what I want to resist, even as I take these developments seriously. One wrong response would be panic, but another would be to dismiss it too quickly.

Marc Watkins’ “nuisance tech” framing is actually a gift, because it gives us a category. Not every alarming AI headline represents a structural shift. Some of it is a 22-year-old with an API key and an attention-seeking marketing strategy. Part of wisdom in a fast-moving technological landscape is developing the discernment to tell the difference between noise and signal.

The signal this week is GPT-5.4. Agentic AI has crossed into the mainstream. It is no longer a feature reserved for developers willing to build elaborate workflows. It is now built into tools your staff may already be using. That is worth understanding clearly before it arrives in your ministry uninvited.

The τέκτων principle applies here as much as anywhere. Jesus, the skilled craftsman and artisan, did not simply pick up whatever tools were lying around and start building. He knew his materials. He understood the properties of stone and how it behaves under pressure. He worked with intention and purpose.

The question for ministry leaders right now is not whether agentic AI is coming, or even whether to engage with it. It is already here, and many ministry leaders I know are already using it quite effectively. The question is whether we are going to engage with it intentionally, with clear policies, theological grounding, and appropriate oversight... or whether we are going to discover its implications the hard way, the way that researcher did when she was sprinting across the room to stop her inbox from disappearing.

Three Questions Worth Asking This Week

1. What does your ministry’s AI agent policy look like? Not AI broadly, but specifically: if a staff member sets up an AI agent with access to church systems, what is it authorized to do? Who is responsible for overseeing it? Do you have a procedure or process in place for fellow staff or IT leaders to review agentic arrangements before they’re deployed? These conversations are worth having now, before an incident forces the issue.

2. Who in your congregation is living inside these disruptions right now? Students, educators, and technology workers in your pews are navigating the Einstein moment in real time. They need pastors who are informed enough to help them think it through theologically, not just technically.

3. What is the difference between nuisance and signal in your context? Develop the habit of asking: Is this alarming because it represents a genuine shift, or because someone built an attention-grabbing wrapper around existing technology? Both deserve a response, but not necessarily the same response. Also, even if this one was more noise than signal, it certainly won’t be the last. There will be more, and they will be far more powerful.

Simply put, the pace is not going to slow down. If this last week’s news means anything, it suggests very clearly that the pace of agentic AI development is rapidly accelerating. The good news is that wisdom is not about being first. It is about being prepared, and preparation starts with paying attention.

That’s why I wrote my book Early Warning, after all…to give pastors and ministry leaders like you a heads-up of what’s coming since early warning systems are game-changing.

I’d love to know your thoughts. Are you already navigating any of these questions in your ministry context? Drop a comment below, or reply directly. These are exactly the conversations τέκτων exists to have.

Resources referenced in this article: Marc Watkins, “Einstein & The Rise of Nuisance Tech,” Rhetorica, February 27, 2026; Michelle Kassorla, “The First Agentic AI LMS Killer is Here,” the academic platypus, 2026; and OpenAI GPT-5.4’s Release Notes, March 5, 2026.

If you are reading this, congratulations! You have found τέκτων (tektōn), a Substack channel I created and recently launched to explore the intersection of technology, innovation, and ministry, with a special focus on the responsible use of generative AI. My goal is not hype nor fear, as neither accomplishes much. Instead, my goal is wise workmanship and careful discernment, responsibly building together for mission.

I had planned a very different inaugural post. But then the world of AI news handed me a story I simply could not set aside. As it turns out, it is also the perfect story to introduce what this channel is about.

Subscribe now

I need to say one more thing before I continue: I almost never wade into political commentary. It is not because I lack opinions (I have plenty), but it is because I know how quickly political framing shuts down real conversation, and real conversation is what I am all about. I loathe divisiveness, and it’s also unbiblical. But what happened this past week is not ultimately about politics. It is about the future of AI, the safety of every person in this country (& the world), and the principles that should guide the technology we use in ministry and in life. That is why I am addressing it.

What Actually Happened

For those who missed it, here is a quick summary. Anthropic, the company behind the AI model Claude, has been working with the U.S. Department of War (formerly known as the Department of Defense) since June 2024. They were, in fact, the first frontier AI company to deploy their model inside the government’s classified networks. By every account, it was a productive relationship. Anthropic supported military intelligence analysis, operational planning, cyber operations, and more.

The dispute arose over two specific requests the Pentagon made: that Anthropic remove safeguards protecting against (1) mass domestic surveillance of American citizens and (2) fully autonomous weapons, meaning weapons that can select and engage targets without a human being in the loop reviewing and authorizing the action.

Anthropic said no.

In a public statement, CEO Dario Amodei explained both refusals in clear terms. On autonomous weapons, Anthropic’s position was straightforward: today’s AI is simply not reliable enough to be trusted with decisions that can end human lives without human oversight. On domestic surveillance, the concern was equally direct: powerful AI makes it possible to assemble scattered, individually harmless pieces of public data into a comprehensive picture of any person’s life, automatically and at massive scale. That crosses a line that violates principles of democracy, and crossing that line was unacceptable.

The Department of War’s response was to label Anthropic a “supply chain risk,” a designation that has historically been reserved for foreign adversaries, and has never before been publicly applied to an American company. President Trump escalated further on Truth Social, directing every federal agency to immediately cease use of Anthropic’s technology. The threats included the full power of the Presidency, major civil and criminal consequences, and – in classic Trump form – degrading characterizations of Anthropic as “woke,” “leftist,” and “out of control” (none of which are true).

Anthropic’s response? They boldly and confidently asserted, “No amount of intimidation or punishment from the Department of War will change our position on mass domestic surveillance or fully autonomous weapons. We will challenge any supply chain risk designation in court.”

Why This Matters So Much

I want to be careful here, because I know this story has already been painted with political colors. Let me be clear: this is NOT a partisan issue; this is a safety and responsible use issue. Millions of conservatives use Claude every day, precisely because it is widely regarded as the safest and most ethically grounded AI model available. It is the LLM that I personally use the most, and I regularly recommend it to educators and ministry leaders, especially when ethics, privacy, safety, student or congregational data, or integrity are important factors.

The concerns Anthropic raised, about surveillance of American citizens without warrants and weapons that kill without human judgment, are concerns that should alarm every person across the political spectrum. The key question is not whether AI should serve national security. Of course it should. In fact, Anthropic was doing exactly that already, and was doing it well. The key question is: which uses are simply off-limits, regardless of who is doing the asking?

That question is at the very heart of responsible AI. And it is why I feel so strongly that this moment deserves our full attention.

Think about what “fully autonomous lethal weapons” actually means. That means a computerized algorithm – a machine – decides to end a human life, with no human in the loop. No human review. That also means there is no human accountability or consequence for decisions made. I want you to sit with that for a moment, because it is no longer a futuristic concern. This used to be something we would only read about in dystopian science fiction, with “worst scale scenarios” of AI that had gone rogue. But that is the actual contract provision being negotiated right now…in America!

Now, to be fair, the Pentagon kept saying they “had no interest” in either and would only pursue “lawful” uses. But why would you terminate an already-active contract and blacklist a company that insisted on those maintaining those provisions if those weren’t your actual intentions?

Anthropic was right to refuse it, not because they are anti-military, but because they understand that AI systems today are not reliable enough to carry that responsibility. Even Dario Amodei acknowledged that fully autonomous weapons may eventually prove necessary for national defense, but the keyword is eventually. We are not there yet. And deploying unreliable systems in that role puts American warfighters and civilians at risk.

Now consider the surveillance piece. Current law allows the government to purchase detailed records of Americans’ movements, web browsing, and associations from public sources, without a warrant. Individually, those pieces of data may seem harmless. But AI can now weave them together into a comprehensive portrait of your life, your habits, your relationships, your beliefs, and your church attendance, automatically and at scale. Our laws have not yet caught up to that reality. Anthropic wisely chose not to help accelerate it.

The Unexpected Twist

Here is something even more intriguing…and possibly disturbing. Within hours of Anthropic losing the Pentagon contract, OpenAI (maker of ChatGPT), stepped in and signed a deal of their own. The reporting indicates that the Department of War agreed to safety red lines similar to what Anthropic had requested, though the leaders involved are widely known for duplicity and double-speak, and government officials simultaneously stated the deal permits all “lawful purposes” without exceptions. Those two descriptions are difficult to reconcile, and the situation warrants continued scrutiny.

But then something else happened. Something remarkable: Claude shot to number one in U.S. app downloads, overtaking ChatGPT for the first time. Social media campaigns encouraging users to cancel ChatGPT and switch to Claude gained tens of thousands of followers almost overnight. A viral video showed chalk art outside Anthropic’s San Francisco offices reading: “You give us courage.”

The market, it seems, has rewarded integrity and responsible use. At least in the short term. Whether that translates into long-term business viability still remains to be seen. But it is a signal worth noting.

What This Means for Ministry Leaders & AI Use

So why am I writing about a Pentagon contract dispute in a ministry technology channel? Because this story is a masterclass in exactly the principles this channel exists to explore.

First, it illustrates why guardrails matter. Not because technology is evil. Not because AI is the enemy. But because powerful tools require thoughtful limits. This is true for hammers, and it is true for AI. As I explore in my book Early Warning, Jesus himself was a τέκτων, a craftsman and artisan who worked with the tools of his day. The question was never whether to use tools. The question was always whether to use them wisely, with purpose, and in ways that served people and honored God.

Second, it illustrates that the companies we choose to trust with our technology matter. Not every AI model is built on the same values. When you choose a platform for ministry use, for sermon prep, for counseling support, for administrative workflow, you are also implicitly aligning with the values embedded in that platform (whether that’s part of your thinking and decision-making process or not). That alignment deserves more attention than most of us give it.

Third, it illustrates that responsible AI use sometimes requires saying no. Just because we can doesn’t always mean that we should. Anthropic said no to a government contract worth potentially hundreds of millions of dollars, because two specific use cases crossed their long-established ethical lines they were not willing to move. That is a costly decision. And it is an example worth studying. Ministry leaders often face similar moments, though not always at that scale. But the question is the same: will we hold the line when it matters most?

Integrity is not about what you say you’ll do. It’s about what you actually do when it costs you dearly. It’s about doing the right thing regardless of the consequences, and regardless of the reputational fallout or financial implications. And courageous leaders know the stakes and live out of conviction, not convenience.

Admittedly, this story is still unfolding. I waited several days to let the dust settle and to make sure I know the full story, but legal battles are likely ahead and the drama will likely continue. The long-term consequences for Anthropic’s business remain unclear. Whether the Pentagon eventually gets its way and whether OpenAI means what it claims are still unknowns that we will have to figure out over time. And the broader question of AI use in military and government contexts is far from resolved. Space does not permit me to explore every dimension of this as deeply as I would like. But I will continue to follow it, and when it matters, I will bring you my analysis as it develops.

Moving Forward

Here is what I know right now. If you are a ministry leader using AI, I want you to feel confident that this channel is going to help you navigate this landscape with clarity and conviction. We are going to talk about what responsible use actually looks like in practice. We are going to look at real tools, real workflows, and real guardrails. We are going to explore the history of how the church has navigated technological disruption before, because we have been down this road before, and there is wisdom in that story.

And when something happens that demands our attention, like this past week, I will bring it to you with honesty, context, and as little political noise as I can manage.

The line held this week. That matters. And so does the work we do together to make sure we build wisely on what comes next.

Welcome to τέκτων.

Sources referenced in this post include Anthropic’s official statements from Dario Amodei (February 26, 2026) and the company’s response to Secretary Hegseth (February 27, 2026), reporting from Axios (March 1, 2026), and the GenAI Works newsletter via LinkedIn (March 2, 2026).

I’m preparing to launch this τέκτων (tektōn) channel by the end of February 2026. I’ve been writing content, gathering resources, and culling some “best of” materials so that I’ve got a healthy supply of great posts ready to share when I officially go live. My plan is to post weekly for everyone, with some bonus posts and sneak peeks for subscribers only.

Subscribe now so that you’ll be guaranteed to receive the inaugural launch edition (tentatively Fri., 2/27). Feel free to pass the “About” page link along to other pastors and ministry leaders you know who might appreciate it.