The Church Is Not Too Slow. It Has Forgotten What Should Stay Slow.

TL;DR — The church is not failing to respond to AI. The Vatican, evangelical institutions, and Taiwan’s church boot camps have all moved faster than we imagine. The problem is not speed but rhythm: members are starting to hand spiritual trust to AI, and pastors are folding AI into the sermon process, but the church has not yet said clearly what may go fast and what must go slow. In the age of AI, the church’s most precious asset is holding on to the capacity for counter-rhythm.

September 2025, Taipei, Shekinah Bread of Life Church. GOOD TV’s “Church AI Boot Camp” opened here: three days, 15 tools, ChatGPT, NotebookLM, Gamma, HeyGen. Registration came with one condition, the senior pastor’s sign-off. Over the following months, Tunghai Bread of Life Church in Taichung and Wuchang Church in Kaohsiung ran boot camps in turn. According to the organizers, the Taipei session alone drew more than eight hundred people.

In those very months while the boot camp toured, the U.S. firm Barna fielded a pastor survey that would be released the following year: 24% of pastors used AI to write or revise sermons, double the figure from early 2024. The Lifeway survey that asked churchgoers found that 43% opposed AI in sermons.

For the past three years, “AI arrived too fast and the church can’t react in time” has probably been the most common anxiety in church circles. I used to think so too. After finishing my research, I found the narrative did not add up.

The church’s institutions are running, and pastors themselves are running.

The real crisis is not slowness. It is that only speed is left.

How fast did the church actually run this time?

First, let me be clear about the subject. The “church” in this piece is at least three different things: the institutional church, meaning the Vatican, denominational assemblies, and church bodies; the individual pastor, that group using AI to look things up and write sermons; and the local church’s pastoral ground, meaning how members use it and whether anyone teaches discernment. The three layers move at different speeds, and blending them only produces argument without result.

Start with the institutional layer, using a method of placing two timelines side by side.

The last time information technology shook faith, it was the printing press. Gutenberg’s movable-type press was running in Mainz around 1450 and printed its first Bible around 1455. Luther posted his Ninety-Five Theses in 1517, and the printing press amplified a scholarly dispute into the Reformation, already 62 years later. I wrote about this stretch in The Real Context of the Reformation. And a systematic counter-response from the church took longer still: the Roman Inquisition in 1542, the first Index of Prohibited Books in 1559, the Council of Trent from 1545 to 1563. The church was not silent for a century; there was a first papal bull on printing in 1487 and general pre-publication censorship by 1515, but the main institutional response took two or three generations from beginning to end.

And AI this time?

ChatGPT went live on November 30, 2022, and analysts estimated it reached about a hundred million monthly active users in roughly two months, the fastest-growing consumer application at the time. Models turn over by the month: GPT-4 (March 2023), Claude 3 (March 2024), GPT-4o (May 2024), Claude 4 (May 2025), GPT-5 (August 2025).

The church’s response, seen within its own history, is abnormally fast. The Southern Baptist Convention’s Ethics and Religious Liberty Commission published Artificial Intelligence: An Evangelical Statement of Principles in April 2019; the Vatican signed the Rome Call for AI Ethics together with Microsoft, IBM, and others in February 2020. Both came before ChatGPT went live (three and a half years and 33 months earlier, respectively). Twenty-six months after ChatGPT went live, the Dicastery for the Doctrine of the Faith and the Dicastery for Culture and Education jointly released Antiqua et Nova, 117 paragraphs, the church’s first doctrinal note dedicated to AI. In May 2025 a new pope was elected, and two days later he explained to the College of Cardinals why he chose the name Leo XIV: because Leo XIII answered the Industrial Revolution with the encyclical Rerum Novarum, and what he must face is another such revolution brought by AI. In May 2026, his first major encyclical appeared, and its theme was AI.

Compared with the printing press, the institutional reaction this time is unusually fast: responses that once took several generations to take shape gradually have already appeared within a few years.

Now the individual pastor. In December 2025, Barna asked 442 pastors: only 13% use AI not at all; half use it to brainstorm, 36% use it to look up biblical and theological material, and 24% use it to write or revise sermons.

So let me say one thing up front: the “fast” in this piece means institutional response and practical adoption have already begun, and does not mean the church has thought things through. The documents came out, the courses opened, the tools are in use.

“Too late” is not without reason, but it is not the most accurate name for this crisis.

The real fracture: a mismatch between trust and use

The survey Barna released (1,514 U.S. adults), fielded in November 2025 and released the following May, asked a question: when it comes to spiritual growth, are you willing to trust AI? 48% of U.S. adults said yes, whether fully or in part.

The same study asked pastors. 12%.

Layer this number over the usage rates from earlier and a strange mismatch appears: pastors do not much trust AI as a source of spiritual growth, yet they have already folded it into the early stages of spiritual labor; churchgoers trust AI four times as much spiritually, yet do not necessarily know what it is they are trusting. In the Lifeway survey, 61% of churchgoers are worried about AI entering faith, and 43% explicitly oppose AI in sermons; and one in four pastors already uses it to write sermons.

Pastors dare not treat AI as a pastor, while members may already be treating it as a spiritual entry point.

Let me be fair here: AI appearing in the sermon process is really several different things. Gathering material, polishing phrasing, and suggesting structure are on a completely different layer from core interpretation and spiritual discernment. The problem was never whether AI appears in the process, but which layer it appears on, and whether anyone has drawn that line.

The default is not neutral: faith gets muted

This makes the question not merely “do people trust AI,” but rather: what default have they actually handed their trust to?

A BYU-led study (1,125 U.S. respondents, released June 2026) offers a clue: most people expect AI to include a religious perspective when answering ethical questions, yet in nearly every model’s default answer, religion is absent.

34% of practicing Christians already think AI’s spiritual advice is as trustworthy as a pastor’s, rising to 39% among Gen Z. Within that same group of practicing Christians, 83% worry AI misreads the Bible and 65% worry AI becomes a substitute for God. Trust has already been given, and the worry is real too.

Layer these figures together and the situation that emerges is this: a group of people, worried even as they do it, hand the entry point of their spiritual life to a system that is silent about faith. At that entry point, faith is not opposed but muted.

Slowness is the church’s craft of two thousand years

I am no stranger to speed.

I am a founder and use AI heavily every day. Models iterate every half month, and my workflow gets rewritten roughly every two weeks. I know this rhythm well, and I genuinely benefit from it. But having lived in that rhythm for a while, I know its cost clearly: in a fast world, nothing has time to mature.

When the World Association for Christian Communication (WACC) discussed AI ethics, Stiven Naatus of the Evangelical Lutheran Church of Finland used an old word from the monastic tradition: acedia, sloth. Its original meaning is not laziness. It points to this: things get finished too quickly, and meaning has no time to mature. He called the contemporary situation a “crisis of rhythm”: people need time to build trust and shape meaning, AI’s speed of output breaks this balance, and so students submit assignments with AI and the public outsources thinking. The role of church liturgy is exactly the opposite, a “counter-rhythm”: slow the pace, preserve space for wisdom and trust to ripen.

Of the church responses I have read, this one comes closest to the heart of the matter. It did not ask “how does the church keep up with AI’s speed.” It asked: why should the church accept AI deciding its speed for it?

Think about what the church holds. Liturgy, meditation, the Sabbath, the tradition of reading a passage of scripture slowly. This is a craft practiced for two thousand years, made precisely to deal with “too fast.” In an age where everyone is forced to accelerate, this craft has for the first time turned from an antique into something scarce. When the whole world is learning how to go fast, the church should at least be one of the few places that still remember how to go slow.

Of course, the church cannot always hold the slow. Much of the time the church itself looks like an activity machine: rushing meetings, rushing courses, rushing headcount, rushing exposure. AI is not the problem itself. It is like a harsh light, exposing problems we had not noticed before. Where is the believer’s Sabbath? Does the believer find rest within their faith, or do they dance to the world’s tempo, filling their life with service after service, just as frantic as everyone else?

Sovereignty over rhythm: what the church should least surrender

I wrote in From 1665 to 2026 that the core question of the age of AI is where the authority to judge actually stays. This piece wants to push one step further: that authority is not abstract. Its most concrete expression today is rhythm.

What can go fast? What must go slow? Who has the right to decide that speed?

That is what I mean by sovereignty over rhythm.

What the church should least surrender is not the right to use the tools, but the authority to judge the rhythm. AI can help organize data, translate documents, draft first passes, and support administration, teaching, and content production. Where speed serves these tasks, let it go fast. There is no need to sanctify slowness for its own sake.

But pastoral work is not only a question of efficiency. Core interpretation, spiritual discernment, how truth is understood: these cannot be pushed into a faster process just because a tool can do it. AI can help organize a sermon, but it cannot replace the pastor’s wrestling with scripture, the congregation, and God. Faith can use tools, but it cannot outsource the time meaning takes to mature.

The real crisis now is that fast and slow have swapped places.

Where things should go fast, they have gone slow instead. Churchgoers are already asking AI faith questions and have begun handing over a portion of their spiritual trust to it. But has the local church taught its members how to discern? Has it made clear which answers can be treated as reference, and which questions should return to the pastor, the community, and the tradition of faith to be judged together? This layer cannot wait any longer.

Where things should go slow, they have sped up instead. One in four pastors already has AI involved in the sermon process, yet most churches have only just begun discussing where the line falls: organizing material is fine, but where does core interpretation end? Polishing language is fine, but can spiritual judgment be handed over? Without these questions answered first, the tools will end up deciding the pace for us without anyone noticing.

So the question is not whether the church should use AI, but whether it can still tell the difference: where it must respond quickly, and where it must insist on going slow.

This is also the question I keep tracking under the theme Civilization and Humanity. The chaser is ultimately replaceable. Tools turn over, courses expire, and of the 15 tools taught today, how many will still be around next year? The church’s irreplaceability lies not in being better than anyone at using tools, but in whether it can still guard, for people, the time it takes for meaning to mature.

In an age where everything is pushed to accelerate, what the church truly needs to hold onto is not the right to fall behind, but the capacity to tell speed apart.

Has the meaning ripened?

A sermon can now be generated in a few minutes.

What that old word acedia reminds us is that whether the generation is good is not the question. Whether the meaning has settled, whether the theological reflection has been thorough, that is the question.

Trust takes time. Comforting someone who has just lost a loved one takes time. This time cannot be saved; what you save is no longer the thing itself.

At this moment when the whole world is accelerating, holding the post of slowness is the church’s calling.