All articles

AI & Ministry

How Many Pastors Use AI? What the Numbers Actually Show

Barna's late-2025 research says 87% of pastors now use AI somewhere in ministry, and about a quarter let it touch the sermon itself. Here's what the numbers actually show—and the tension inside them.

Jon Horton · August 3, 2026 · 3 min read

If you've quietly wondered whether other pastors are using AI, here's the answer: yes, almost all of them, and almost none of them feel settled about it.

The most useful data comes from Barna's research on pastors and AI, released in late 2025 from a survey of over four hundred U.S. pastors. Three numbers tell the story.

The three numbers

87% of pastors use AI somewhere in their ministry. Emails, planning, research, graphics, admin. Only about one pastor in eight reports using no AI at all. Whatever the debate sounds like online, the adoption question is settled—AI is already in the church office.

About 24% use AI to write or edit sermons. This is the number that matters for preaching. Roughly a quarter of pastors let AI touch the sermon itself—up sharply from the year before, when comfort with AI in sermon work sat near half that. The trend line points one direction.

79% carry some fear about AI replacing what only God should do. Read that against the first number and you have the honest picture: the overwhelming majority of pastors are using a technology that the overwhelming majority of pastors are uneasy about. Adoption did not wait for peace of mind.

What the numbers don't say

A survey checkbox called "uses AI to write or edit sermons" flattens the distinction that actually matters. Inside that 24% are pastors doing very different things:

  • Some ask AI background questions during study—the digital equivalent of pulling a commentary off the shelf.
  • Some use it to tighten sentences in a draft they wrote themselves.
  • Some generate outlines and fill them in.
  • Some paste in a passage Thursday night and preach the output Sunday.

The first two describe help. The last describes replacement. The survey can't tell them apart, and neither can the congregation—which is exactly why the interesting questions aren't statistical. They're the ones every pastor in that 87% eventually has to answer personally: where's the line, is crossing it cheating, and would you say out loud which side of it you prepare on?

The tension is the finding

Here's our read of the whole dataset in one sentence: pastors have concluded AI is too useful to refuse and too dangerous to trust, and they're holding both conclusions at once, mostly alone, mostly without telling anyone.

That's an uncomfortable place to stand, but it's not an unreasonable one. Both instincts are correct. The usefulness is real—the pastor who gets three hours back in a week gives them to people. The danger is real too—the wrestling with the text is where God works on the preacher, and tools now exist that will quietly remove it.

The resolution isn't picking an instinct. It's picking a line—deciding in advance what you'll accept help with and what has to stay yours—so that adoption becomes a conviction instead of a drift. That's the entire design premise behind the coaching category of tools, ours included: all of the help, none of the substitution, so the number you're part of never has to bother your conscience.

One step: if you're in the 87%, write down—one sentence—what you currently let AI do in your prep. Just look at it. Most pastors have never actually stated their own practice out loud, and naming it is the fastest way to find out whether you've drawn a line or drifted past one.