The conversation about AI for pastors has mostly been conducted at the extremes — either it is the end of authentic preaching or it is a miracle that gives you Saturday back. The research published this summer suggests the reality in most church offices is quieter and more interesting than either.
In December 2025, Barna Group surveyed 442 U.S. Protestant pastors for a study with Gloo, released in June 2026. Eighty-seven percent said they use AI in some part of ministry. In the same survey, seventy-one percent described their feelings toward it as “cautious,” forty percent as “conflicted,” and forty percent as “skeptical.” In other words, the typical American pastor is already using these tools and is not sure they should be.
This guide is for that pastor. It covers what the data actually says, what AI is good and bad at in the specific work of ministry, a table of uses we would and would not recommend, a short theological frame, and a way to talk to your congregation about it.
Key takeaways
- 87% of U.S. Protestant pastors use AI somewhere in ministry (Barna/Gloo, 2026). Only 24% use it for sermon writing or editing; the main uses are brainstorming and research.
- Pastors are far more cautious than the people in their pews: 71% of pastors say “cautious,” while 34% of U.S. Christians say “hopeful.”
- AI is strong at gathering, organizing and drafting and weak at accuracy without verification, specificity and discernment. Build your practice around that.
- The line most pastors draw — and the one we recommend — is between research assistance and ghostwriting. Everything you preach should be decided and written by you.
- Tell your church how you use it. Silence invites suspicion; a plain policy invites trust.
What the 2026 research actually says
Two numbers from the Barna/Gloo study get quoted; the rest of the table is more useful.
| Use of AI in ministry | Share of U.S. Protestant pastors |
|---|---|
| Brainstorming and idea generation | 50% |
| Graphic design and visual creation | 37% |
| Biblical and theological research | 36% |
| Writing small-group discussion questions | 34% |
| Administrative tasks | 34% |
| Sermon writing or editing | 24% |
| Do not use AI at all | 13% |
Source: Barna Group / Gloo, U.S. Protestant pastors, n=442, December 2025.
Three things stand out. First, the most common use is the least controversial: brainstorming. Second, the most-discussed use — sermon writing — is the least common, though it has doubled from 12% in early 2024 to 24% now. Third, “research” at 36% is the number to watch, because it is where the accuracy problems live (more on that below).
The emotional data is just as telling. Asked how they feel about AI, pastors said cautious (71%), curious (52%), conflicted (40%), skeptical (40%), hopeful (18%) and excited (10%). Practicing Christians in the parallel adult survey were markedly warmer — 34% hopeful and 32% excited. The gap between the pulpit and the pew is one of the reasons we suggest saying something to your congregation rather than nothing.
Finally, the concerns: 79% of pastors worry that AI could “act as a replacement for God” in people’s lives, and 63% worry about AI replacing pastoral roles. Those are not fringe anxieties. They are the majority view of people who are, at the same time, using the tools. We think the right response is neither to dismiss the worry nor to abandon the tool, but to use it in a way that keeps the pastor in the pastor’s seat. We look at the full dataset in How Pastors Are Using AI in 2026.
What AI is good at in ministry work
Be specific about strengths and you will use the tool for the right jobs.
Gathering. Given a passage, a good AI tool can assemble historical background, the place of the passage in the book, the interpretive options, and cross-references in a minute. This is the work that used to take an hour with two commentaries — the “mechanical” part of study.
Organizing. AI is unusually good at turning a page of your own messy observations into a structure: “here are the three moves in this text, and here is the verse that supports each.” The output is better when the input is yours.
Drafting. First drafts of discussion questions, announcements, a newsletter paragraph, a Series Plan for Advent, or the opening two sentences you cannot get right at 11 p.m.
Asking questions back. The best use of an AI chat in study is not “write me a sermon” but “what would a skeptical listener object to in this point?” or “which interpretations of this verse have serious scholarly support?” — questions a good study partner would ask.
Volume tasks nobody has time for. Small-group questions for every week of a series; alt text for every image on the website; a first pass at a grant letter. This is where the 34% doing admin with AI are quietly saving real hours.
What AI is bad at — and how to protect against it
It fabricates. Large language models will produce a Bible reference, a quotation from Spurgeon, or a statistic with total confidence and no basis. In sermon work this is the cardinal risk. The rule is simple: no reference, quote or number goes into a sermon unless you have opened the source yourself. Tools that ground their output in the actual Bible text reduce this problem but do not eliminate it. Our article on AI hallucinations and Bible references gives a four-step verification routine.
It is generic by default. Ask for an illustration of grace and you will get the same three stories every pastor with the same tool gets. Ask for an illustration of grace for a congregation of dairy farmers in a county that lost its hospital, and you will get something you could not have found in a database. Specificity in, specificity out.
It does not know your church. It has no idea who is in the third row, whose marriage is failing, which family buried a child in March. Application is where a sermon meets those people, and it is the part of preparation that cannot be delegated.
It flattens theology. Models are trained on everything and will happily blend traditions. If you preach from a Reformed, Wesleyan, Baptist or Pentecostal conviction, you will need to tell the tool that and still check the output.
It makes easy things too easy. The subtlest danger. A structure that arrives in thirty seconds is easy to accept, and accepting it means skipping the hour of reading that would have produced a better one — and would have formed you. Any tool that shortens preparation has to be paired with a discipline that protects study.
A use-or-don’t table for pastors
| Task | Our recommendation | Why |
|---|---|---|
| Historical and literary background for a passage | Use — then verify key claims | High value, low risk when checked |
| Original-language word notes | Use with care | Check against a real lexicon; beware root fallacies |
| Proposing a sermon structure from your notes | Use | Removes the blank page; you still decide |
| Generating the sermon from a passage with no input | Don’t | Produces a generic sermon and skips your formation |
| Illustration shortlist | Use — with your context supplied | Biblical/historical suggestions are strong; contemporary ones weak |
| Application | Draft only | Rewrite for the actual people in your room |
| Finding a quotation or statistic | Don’t trust; do verify | Fabrication risk is highest here |
| Small-group questions | Use | Time-saver with low downside |
| Bulletin, announcements, newsletter copy | Use | Admin — nobody’s calling is at stake |
| Counseling preparation (scenarios, passages, questions) | Use for preparation only | Never a substitute for the pastor or a licensed professional |
| Pastoral care messages to individuals | Don’t | People can tell, and it matters that they can |
| Writing your church’s AI policy | Use as a starting draft | See our template |
The line most pastors draw
Across the pastors we have read and spoken with, one distinction keeps appearing, and we think it is the right one: research assistance versus ghostwriting.
A concordance, a commentary, Bible software, a seminary lecture you took notes on — all of these have always been legitimate sources of sermon material, provided the pastor did the work of understanding, deciding and writing. AI used the same way sits comfortably in that tradition. Ghostwriting — preaching a message you did not decide on, in words you did not write, as though you had — has never been acceptable, whether the ghost was a book, a website or a model. The technology is new; the ethical category is not.
That gives a practical test for any AI output: Could I explain and defend every sentence of this to my congregation as my own considered conviction? If yes, it has become yours. If not, it has not, however good it sounds. We develop this argument, including the objections about integrity and dependence on the Spirit, in Is It Wrong to Use AI for Sermons? and the plagiarism question in AI, Plagiarism and the Pulpit.
Choosing tools: general assistants vs. purpose-built
There are two kinds of AI tool a pastor can use, and the choice matters more than the brand.
General assistants (ChatGPT, Claude, Gemini and similar) are inexpensive, flexible and improving quickly. Their weakness is that you must design the process yourself — decide what to ask, in what order, and check every reference — and they are not grounded in a Bible text by default. Used well, with a set of saved prompts, they are a fine research partner; we give fifteen such prompts in ChatGPT for Sermon Preparation.
Purpose-built tools build the process in. Pastor Rhema, for example, walks a pastor through Study & Context, Sermon Structure, Illustrations, Applications and Final Sermon in that order — the same five stages we describe in how to prepare a sermon — and exports a draft to Word, PDF or slides; other tools take different approaches at different prices. The advantage is that you cannot skip a step and the output is organized; the cost is a subscription and less flexibility than a blank chat. Our comparison of the best AI tools for pastors covers nine of them with prices verified on the day of publication.
Whichever you choose, the verification rule and the ownership test above apply equally.
Talking to your congregation about it
The Barna data gives you the argument: your people are, on average, more hopeful about AI than you are. Silence, on the other hand, lets the 24% figure — “pastors are using AI to write sermons” — become the story someone tells about you.
A short statement does most of the work. Three tiers, from lightest to fullest:
- A line in the church FAQ or newsletter: “Like most pastors, I use AI tools for research and organization in sermon preparation. The study, the decisions and the writing are mine.”
- A paragraph in a members’ meeting, with the research-vs-ghostwriting distinction and an invitation to ask.
- A written church AI policy covering staff use, disclosure, confidentiality (never put a congregant’s private situation into a public AI tool), theological review and accountability. We provide clauses and sample language in our church AI policy template, and a fuller disclosure framework in Should You Tell Your Congregation You Use AI?.
A pastoral word about time
There is a reason 87% of pastors have reached for these tools, and it is not laziness. Lifeway’s research has long shown that most pastors spend eight or more hours a week on sermon preparation alone, and Barna’s well-being studies have tracked a rise in pastors considering leaving ministry — 38% in one 2021 survey, driven by stress, loneliness and workload. A tool that returns two hours a week to a pastor who spends them with a sick parishioner or with their own children is not a threat to ministry. Used carelessly, the same tool can hollow out the study that forms a preacher. The difference is not the tool. It is the discipline around it — which is what this guide has tried to give you.
Conclusion
AI for pastors is no longer a question of whether but of how. The 2026 research shows a profession that has adopted the tools and remains uneasy about them, and we think that unease is healthy as long as it produces discipline rather than denial. Use AI to gather, organize and draft. Verify everything. Supply your own context. Keep the decisions and the writing yours. Tell your church what you do. And protect the hours of reading and prayer that no tool can do for you — because those hours are not the mechanical part of preparation. They are the ministry.
