What the Printing Press Teaches Us About AI and the Church
The church has already lived through a technology that made truth and error travel at the same speed. It did not go the way you would guess.
Gutenberg’s press printed two things in its first years: a Bible, and indulgences. Same shop, same movable type, same astonishing new ability to run off a thousand identical copies of anything. One of those products put Scripture into more hands than any technology in history. The other industrialized the sale of forgiveness, and within a lifetime it helped blow the Western church apart.
I keep returning to that pairing, because we are standing in the same kind of moment. A general-purpose technology has arrived, it is cheap, it is spreading faster than anyone’s policy for it, and it is entirely indifferent to what we point it at. The printing press is the closest thing we have to a rehearsal — and the church has already run this experiment once, at enormous cost, and learned things worth not relearning the hard way.
Key takeaways
- The press spread Scripture and error at exactly the same speed. Access and distortion scale together — always.
- The church’s first instinct was control: licensing, bans, burnings. It lost, and it lost quickly.
- What actually worked was formation — catechisms, study notes, teaching people to read well, not merely to read.
- The press never replaced the preacher or the congregation. Neither will AI. Know what the tool can’t touch.
What actually happened after Gutenberg
The speed is the part we forget. Within fifty years of that first Bible, presses were running in more than two hundred European towns, and historians estimate they had produced somewhere around twenty million volumes — more books in half a century than Europe’s copyists had turned out in the previous thousand years. A text that once cost a scribe a year of his life could now be had for the price of a few days’ wages.
The results were glorious. Erasmus published a printed Greek New Testament in 1516, and suddenly scholars across the continent could argue from the same text. Luther’s German New Testament appeared in 1522 and sold out so fast the printers could barely keep up. William Tyndale’s English New Testament was printed on the continent in 1526 and smuggled into England hidden in bales of cloth, so that — in his famous hope — a boy driving a plow might know more of Scripture than the clergy did.
And the results were also ugly. The same presses poured out pamphlets full of libel and invective. The Malleus Maleficarum, a manual for hunting witches, became one of the era’s reliable bestsellers and fed decades of killing. Printed indulgences did a brisk business precisely because printing made them cheap to produce. The machine did not know the difference between the Gospel of John and a lie about your neighbor, and it moved both at the same velocity.
This is the first lesson, and it is not a comfortable one. Every technology that lowers the cost of distributing truth lowers the cost of distributing falsehood by the identical amount. AI is doing this again right now: it makes a grounded, well-cited answer about Habakkuk nearly free to produce, and it makes a confident, fabricated verse about Habakkuk exactly as free.
The first instinct was control — and it failed
Faced with all this, the institutional church reached for the obvious lever. Printers were licensed. Books were burned. Translations were declared unlawful, and in 1559 Rome published a formal Index of prohibited books. Tyndale himself was hunted down, strangled, and burned in 1536 for the crime of putting Scripture into ordinary English.
Here is the detail I find impossible to shake: three years after Tyndale’s execution, an English Bible drawing heavily on his own translation was ordered placed in every parish church in England. Three years. The suppression didn’t merely fail on principle — it failed on the calendar, and the men enforcing it lived to see the reversal.
It failed because bans lose to economics. Once a technology makes something a hundred times cheaper, prohibiting it means asking thousands of independent operators to leave money on the table forever. That was never going to hold in the sixteenth century, and a policy of “our people simply won’t use AI” will not hold now. Your members are already asking a chatbot the questions they are too embarrassed to bring to you. The only live question is whether anyone has helped them ask well.
What worked was formation, not prohibition
The church’s durable answer to the press wasn’t suppression. It was teaching.
Luther wrote his Small Catechism in 1529 because he had visited the parishes and been horrified by how little ordinary people knew — and print let him put a compact, memorizable frame for the faith into thousands of households. The Geneva Bible of 1560 shipped with marginal notes, cross-references, and study helps, because its editors understood that handing someone a text is not the same as equipping them to read it. Catechism, commentary, and study aid are all, in a sense, technologies invented to answer a technology.
That is the move to copy. Not “don’t use AI,” but “here is how you check it.” Not a ban from the pulpit, but ten minutes in a small group actually opening the tool, asking it a doctrinal question, and looking up every reference it produces together. Discernment has never been transmitted by prohibition. It is taught, slowly, by people who model it.
Four things the press teaches us about AI
1. Assume error scales with access. Whatever your tool makes easy, it makes easy in both directions. Prefer systems that retrieve and cite real sources over systems that generate fluent text from memory, and check the citations often enough to know which kind you are holding.
2. Watch what the technology monetizes. The press printed indulgences because indulgences sold. It’s worth asking, plainly, what a given AI product is built to maximize — your understanding, or your time in the app. The answer shapes everything downstream.
3. The vernacular question is back. The Reformation ran on the shock of hearing Scripture in the language you actually think in. Roughly half the world still has no serious digital study resources in its own tongue. Whoever ships in a language first will shape how faith is spoken there for a generation — which is a responsibility, not a market opportunity.
4. Speed is not the same as depth. Print made access instant and left formation exactly as slow as it had always been. AI is doing this again, more sharply. Getting an answer in four seconds does nothing to shorten the years it takes for that answer to become character.
What the press never touched
For all the upheaval, some things were untouched. Printing didn’t replace the preacher, and it didn’t replace the congregation. Nobody was ever discipled by a warehouse. The book could put the text in your hand; it could not sit with you after a diagnosis, or know your name, or serve you communion.
“The grass withers, the flower fades, but the word of our God will stand forever.” — Isaiah 40:8
Isaiah wasn’t talking about media formats, but the principle carries. Scroll, codex, printed page, glowing rectangle, model weights — every container we’ve invented for the word of God has eventually withered, and the word has kept standing. That should make us both bolder and calmer about new tools than we usually manage to be.
It’s the standard we hold ourselves to at SoapBox. ORA, our study companion, answers from Scripture it actually retrieves and cites, in 140+ languages, so you can tap a reference and read it in full — and when a question can’t be grounded, saying so is the honest response. The press was neither the church’s savior nor its enemy; it was a lever, and what mattered was who leaned on it and why. AI is the same kind of lever. We get to decide, again, what we print with it.
Explore faith at your own pace.
SoapBox is a free Bible app with a live prayer wall, daily devotionals, and ORA, an AI study companion that answers questions about faith with grounded, Scripture-cited responses — judgment-free, in 140+ languages. There's a private mode if you're just exploring.
Frequently asked questions
Is AI really comparable to the printing press for the church?
The useful comparison isn’t about scale but about mechanism. Both technologies dramatically lower the cost of producing and distributing religious content, and both are indifferent to whether that content is true. The press spread vernacular Bibles and witch-hunting manuals with equal efficiency; AI produces grounded, cited answers and confidently fabricated ones at the same speed. That symmetry — access and error scaling together — is what makes the history worth studying.
Should churches ban AI or teach members to use it?
History strongly favors teaching. The church’s attempts to control the printing press through licensing, book burnings, and prohibited-book lists failed within decades, because bans lose to economics once a technology makes something far cheaper. What endured instead was formation — catechisms, study notes, and marginal references that taught people to read well. The equivalent today is teaching discernment: how to check a citation, and when to bring a question to a person instead.
What did the printing press actually change about how people read the Bible?
It moved Scripture from something you heard read to you into something you could own, mark up, and check for yourself. Erasmus’s printed Greek New Testament in 1516 let scholars argue from a shared text; Luther’s German New Testament in 1522 and Tyndale’s English one in 1526 put it in the language people actually thought in. That shift also created a new need — study aids and catechisms — because handing someone a text is not the same as equipping them to read it.
How can I tell whether an AI answer about faith is trustworthy?
Ask whether it cites something specific you can open — a book, chapter, and verse rather than “the Bible teaches” — and then actually open it a few times to see whether the passage exists and says what was claimed. Ask what corpus and translation it draws from, and whether it distinguishes what the text says from how Christians have interpreted it. A tool that won’t tell you where its answers come from is asking for trust it hasn’t earned.