Pope Leo and Sam Altman agreed the judgment stays human. Churchgoers are less sure.

What each of them actually said

Leo was not arguing that the technology is bad. He was arguing that there is a category difference between a person making something and a system assembling it.

"There is an ontological difference, even before an aesthetic one, between art and what a machine can generate." A machine's output, he said, comes from "statistical calculation based on millions of images created by others." Then the line that travelled furthest: "Algorithms lack the spark of the human."

Altman's post is not about art at all. It is about what people are doing with the models his company builds, and the phrase worth slowing down on is "a surrender of human judgment." He is not objecting to anyone finding the technology impressive. He is objecting to people handing it the authority to decide.

One is a claim about what a machine is. The other is a claim about what people are doing with it. They arrive at the same practical instruction from opposite ends: the judgment stays with you.

Then somebody checked

On 22 October, Barna published research asking a narrower and much more awkward version of the same question. Not "is AI dangerous," but: at what point should a chatbot stop giving advice and put you in front of a human being?

They asked pastors and they asked practising Christians, and on most situations the two groups broadly agree. Where someone is at risk of harming themselves, almost every pastor says hand off immediately, and so do most churchgoers.

Then there is the spiritual crisis.

Seven in ten pastors say that is a moment to stop and find a person. Barely three in ten practising Christians agree.

Which is to say: on the one crisis that is explicitly the church's own business, the people in the pews are more comfortable letting the machine carry on than the people in the pulpit are.

The people in the pews are more comfortable letting the machine carry on than the people in the pulpit are.

On Barna's spiritual-crisis finding

Why that gap is interesting rather than alarming

The easy reading is that congregations have been quietly replacing their pastors with a chatbot. That is almost certainly not what is happening.

A likelier reading is availability. A spiritual crisis rarely announces itself during office hours. If the question arrives at one in the morning, about doubt, or guilt, or whether any of this is true, the chatbot is there and the pastor is asleep.

The human handoff only works if there's someone to hand off to.

That is Daniel Copeland, Barna's vice president of research, putting the problem in one sentence.

Which is where this reads differently from here

Most church guidance on artificial intelligence is written for an institution choosing between doing the work itself and handing it over. That is a real choice when there is a budget and a team.

For a great many congregations in India there is no such choice. There is one volunteer with a day job, or there is nothing at all. The question of whether to delegate does not arise, because the alternative to the tool is not a person doing it better.

Copeland's sentence was written about American churches with staff. Read it again from a church of forty people in Koraput and it stops being a caution and becomes a description.

None of which makes the advice wrong. Leo's point is that the spark belongs to the person, and in a church with one volunteer that person is unmistakably present. Altman's point is that the judgment stays human, and in a small congregation the judgment is sitting right there in the room, usually with a phone and a hymn list to sort out before Sunday.

Nobody in the Consistory Hall was picturing a parish in Koraput. The advice holds anyway. It is the handoff that does not.

Barna surveyed 1,514 US adults in November 2025, including 683 practising Christians, and 442 US Protestant pastors in December 2025. Margins of error 2.0% and 2.8%.