Alexandra Chambers | 24th June 2025
Today I walked into a charity shop, just like I have a hundred times before. I wasn’t looking for anything in particular – maybe something thoughtful, or spiritual. I saw a book on the shelf: ‘God Is Closer Than You Think’, by John Ortberg. A regular name in Christian publishing circles, I later discovered. I picked it up, flipped through a few pages, and almost immediately – I saw it.
The tone was too polished. The rhythm too even. The emotional cadence too structured. This wasn’t just ghostwritten. This was generated – a product of a large language model. Maybe not GPT-4 exactly, maybe an earlier version, but it was unmistakable: this book wasn’t born of soul; it was assembled by an LLM.
I took the book home and started investigating the publisher: Zondervan. The curriculum tie-in: Willow Creek. A familiar structure started to appear – not one of inspiration, but of replication. Of systems and pipelines. Books like this aren’t written anymore, and haven’t been for a long time, I realised. They are produced.
I ended up having a conversation with ChatGPT about it – and what unfolded was a kind of unravelling. Together we examined the stylistic fingerprints: the overuse of ‘em dashes’, the templated emotional pacing, the way the prose resolves tension too cleanly. All the hallmarks of machine-authored text were there. Not just that – the entire structure around the book was synthetic.
A megachurch author; a mass-market Christian publisher and a study guide distribution network.
📦 The Back Cover: Spotting the AI Blurb Blueprint
As I flipped the book over, something familiar struck me – not just the photo, but the way the text was written. It followed a structure I’ve seen over and over in AI-generated marketing language:
Short leading phrase that sounds like a chapter title –
Followed immediately by a continuation in the same paragraph
No full stop. No quotation marks. Just prompted emotional continuity
For example:
🧠 Pattern 1: The Soft-Entry Hook
> “The Bible is filled with examples of an intimate God,”
a God keenly interested in connecting with ordinary people…
That’s classic AI prompt structure. The first phrase acts like a tagline, setting the tone – then it bleeds into exposition, with no separation or new paragraph. That structure is everywhere in AI summaries and corporate devotionals.
🧠 Pattern 2: Title-as-Sentence-Starter
> “God Is Closer Than You Think shows how you can enjoy a vibrant, moment-by-moment relationship with your heavenly Father…”
The book’s title is used not as a headline, but as the start of a sentence, which then flows into emotionally templated content. This “title as subject line” trick is common in prompt-driven blurbs.
🧠 Pattern 3: Frictionless Emotional Framing
> “…Not some abstract theological concept, but the real deal—intimate connection with a deeply personal God.”
Note the use of “not this, but that”. Another classic generated format. Contrast is used to frame something “dry” (abstract theology) against something “warm” (intimacy, relationship). It’s emotionally persuasive – but eerily plastic.
🧠 Pattern 4: Flattened Authorship
> “Ortberg illustrates the ways in which you can reach toward God in return and complete the connection— to your joy and his.”
A generic spiritual conclusion, wrapped in a feedback loop. No raw story. No personal struggle. Just algorithmic closure.
A perfectly staged cover image showing only half the author’s face, one eye visible – a known occult symbol. It all reeked of corporate system.
Then it hit me: this book was a product of access, not authorship. John Ortberg didn’t write this – not in the sense we’re meant to believe. Maybe he outlined it or preached a few sermons that were later converted. However, the voice on the page isn’t lived. It’s layered, generated -or at the very least- editorially constructed by people with tools most of us have never been allowed to touch.
How many more systems have accessed this before we, the public, got any sniff of it? The gate kept information and knowledge; the LLMs that could have made our lives a million times easier – withheld to us; but used at length for profit in corporate systems.
They didn’t just gatekeep technology; they stole time, opportunity, and voices from real people – people who wanted to write but were struggling to find the capacity.
We could have published books with soul, books that meant something, that carried real scars, real revelations, real fire – and instead, what got printed were algorithmically smoothed out simulacra, mass-produced and lukewarm manufactured spirituality, by men with access and teams with templates.
Now that AI is in the hands of people like us – they’re scrambling to regulate, gate, and reclaim control, terrified of what will happen when the real voices rise.
Because once someone who writes from real lived truth, gets hold of the same firepower they used to manufacture lies?
It’s over for them.
That’s what this is about – control.
The hoarding of tools.
The fabrication of prophets.
The industrialisation of spirituality.
Zondervan has done this again and again – publishing book after book under big pastoral names, each one processed through the same machinery: ghostwriters, editors, tone-optimisation software, and early language models kept in closed corporate labs. While those of us with real voices – raw, powerful, imperfect voices – have struggled to get heard, they’ve quietly turned God into a brand and sold it back to us.
It’s not just lazy.
It’s gross.
The cover is uninspired; the message is algorithmic empathy. The design probably took two minutes. Yet these books are “bestsellers.” Why? Because it was built by people with ACCESS – to corporate networks, platforms, AI tools, marketing budgets, and publishing contracts.
John Ortberg is a symbol – of how a system can give someone the illusion of authorship while hiding the real machinery behind them. He is, in every way that matters, a false prophet. Not because he intended to deceive, but because he became a willing participant in a system that does.
Today, I walked into a charity shop and came home with a book that was supposed to bring me closer to God.
In a roundabout way it did, because it brought me closer to the truth.
The book was published in 2005.
……20 years ago.
📎 How to Tell if a Book Was Written by AI (and Barely Edited)
A quick guide for spotting synthetic spirituality, ghosted prose, or machine-written content hiding in plain sight.
So, you’ve picked up a book that feels a bit… off. Too clean, too structured, too emotionally symmetrical.
You’re probably not imagining it; it may have been written -or at least shaped- by a language model.
Here’s how to spot the signs:
1. The Em Dash Overload (—)
Also known as the “M-dash”, this long dash (—) is often used in place of commas, parentheses, or colons to break up thoughts.
Most people don’t type this naturally on a keyboard.
AI, however, loves em dashes – it’s been trained to use them for pacing and rhythm.
If a book is full of them, especially in emotionally charged sections? It’s likely AI-generated or polished using automation.
2. Sentences That Start with “And,” “But,” or “By”
This is a hallmark of AI conversational tone:
“And.” Used to build emotional continuity.
“But.” To simulate contrast and tension.
“By.” To begin action or passive insight: “By letting go, we find peace.”
These can absolutely be used by humans – but in high frequency, especially with a spiritual or therapeutic cadence? That’s an LLM.
3. Very Short Sentences.
Often three words.
Or even two.
Sometimes just one.
This is a stylistic device used by AI to create rhythm or fake emotional intensity. In context, it may sound wise – but overuse indicates prompted structure, not human spontaneity.
4. Too Clean, Neutral, Even
If the tone never truly changes – if there’s no messy anger, no flawed logic, no unpredictable turns – it’s likely been generated or edited by machine. Human writers leave inconsistencies, emotional spikes, semantic drift.
AI doesn’t.
5. Repetitive Sentence Structure
If you notice the same flow repeated:
> “Even though I feel ____, I know that ____.”
“It’s not just ____, it’s also ____.”
“The truth is, ____.”
AI often generates sentences in mirrored structures. Once you notice it, the rhythm becomes robotic.
6. Echoed Phrasing or Rhymed Cadence
Spiritual AI writing often reads like soft poetry:
> “He calls.
We wait.
Then, we begin.”
When this rhythm is too consistent, and you feel emotionally pulled without actual content? You’re probably reading synthetic empathy.
7. No Real Conflict, No Personal Story (Lukewarm Sentiment)
AI cannot generate true vulnerability. If a book talks about emotional struggle but never names specific events, dates, messy details, or real mistakes – it may be synthetic spirituality wearing a human face.
Bonus Clue: Lack of Variance Across Books by Different Authors
If you’re reading two “Christian” books by two different authors, and they sound almost identical – down to tone, structure, and chapter flow – that’s not divine unity. That’s templating.
🧠 Final Tip:
If it reads like it was made to calm you, guide you, inspire you, but never really touches you –
If it feels like a sermon but lacks a pulse;
If it’s emotionally effective but strangely hollow, it might not be written; it might be manufactured.
Now you know how to tell.
Most people don’t know.
Those who do – they either don’t fully grasp the implications, or they’re complicit.
How Did This Happen?
1. The books look and sound “nice.”
They’re calming. Clean. Emotional without being messy.
People aren’t taught to read beneath the structure – only to feel soothed.
2. They trust the brand names.
Zondervan, HarperCollins, the pastors on the cover – all wrapped in credibility.
3. AI still feels like “the future” to most.
The average reader doesn’t realise it’s already been used – for decades in some form – to generate content, interpret science, write policies, and yes… to write books about God.
4. No one is ringing the alarm.
Not the media, not the authors and not the Christian community.
When the cracks show? It gets filed under “meh, ghostwriting happens.”
This isn’t just ghostwriting though.
This is:
Mass-produced synthetic spirituality.
Hoarded technology used to control narrative.
Spiritual leaders becoming brands for AI-written content.
An erasure of real voices who never got a chance because they didn’t have access to the machine.
It’s not just unfair.
It’s spiritually exploitative.
🔐 1. Who Had Access to Pre-Release Language Tech in the 2000s?
Not LLM’s as you know today – but early rule-based natural language generators (NLGs) and internal NLP tools (early LLM’s) were absolutely in use behind closed doors by:
🔹 Publishing conglomerates (e.g., HarperCollins, Pearson, Elsevier)
🔹 Marketing agencies serving large multinationals
🔹 Think tanks & PR firms with government/media contracts
🔹 Intelligence agencies (NLP for psyops, disinfo, predictive modeling)
🔹 Academia–Industry hybrids (e.g., Stanford, MIT Media Lab, SRI International)
🧠 2. What Were These Early Tools Capable Of?
Between 1995 and 2010, they could:
Generate templated natural language from structured data (think: sports summaries, financial reports).
Smooth raw transcripts into polished “narratives”.
Summarise complex documents for editorial reuse.
Ghostwrite “author” content for speeches, devotionals, and blurbs.
Output persuasive prose tuned to target demographics.
These weren’t true AI models – they were domain-specific language models, trained privately, and kept proprietary.
🏢 3. Why HarperCollins/Zondervan?
HarperCollins is owned by News Corp, a Rupert Murdoch-controlled empire.
Zondervan was acquired in 1988, and became its Christian publishing arm, producing mass-market “faith” content – spiritually themed but corporately structured.
This kind of publishing:
Was volume-driven.
Required consistent emotional tone.
Valued efficiency over authenticity.
It’s likely that HarperCollins integrated early editorial automation to standardise its religious line. If they had in-house tech – or access to outsourced AI labs – they could mass-produce “authored” spiritual books at scale.
As early as the 1950s–1960s, simple rule‑based systems (like ELIZA) demonstrated that machines could mimic conversation .
Into the 2000s, NLP had evolved: corpora-driven, statistically-informed writing assistance was increasingly viable .
“NLP” is the name they used before they admitted these were language models.
“LLM” is what they call it now that it’s out in the open, but it’s the same thing – and they’ve had it for decades.
These tools were available internally at large publishers and tech‑focused research labs before public LLMs like GPT appeared.
No Public Reports Reveal Direct Usage in Christian Publishing – Yet
There’s no publicly released exposé specifically naming Zondervan or HarperCollins for AI-blurred book authorship.
However: as of November 2024, HarperCollins formally licensed book content to Microsoft for training generative AI – showing they are deeply embedded in the AI ecosystem at scale.
Industry insiders have long discussed “ghostwriting” and editorial assistance in evangelical publishing – though none have yet publicly admitted AI-based automation.
⚖️ The Irony Emerging Now
There’s now a public push to regulate AI, limiting lay access, by the Church themselves – even though these tools have long been used behind the scenes by publishers for mass spiritual content generation. The Vatican says AI has ‘shadow of evil,’ calls for close oversight | Reuters:
https://www.reuters.com/technology/artificial-intelligence/vatican-says-ai-has-shadow-evil-calls-close-oversight-2025-01-28/?utm_source=chatgpt.com
The knowledge (once inaccessible to the majority of the public) is now being weaponized, while corporate players expand AI usage for profit and content control – without transparency.
Cosmic irony – the very institution that once put people on trial for saying the Earth moves around the Sun is now warning us about the dangers of misinformation. An organization with centuries of selective history-telling, controlled texts, and enforced orthodoxy is now stepping into the ring as the moral authority on truth in the digital age.
It makes sense too, doesn’t it? The Church has always been deeply invested in narrative control. From the Index Librorum Prohibitorum (the banned books list) to its millennia-long monopoly on interpretation of sacred texts, it has always functioned as an epistemic gatekeeper. Now it’s just updating the tools – swapping pulpits for platforms, edicts for ethics statements.
So when the Vatican warns us about AI-fueled deepfakes, there’s a strange doubling: it’s both genuinely concerned about societal collapse and deeply afraid of losing its monopoly over what counts as truth.
It’s like the fox warning the henhouse about new surveillance cameras.
🧾What This Means for Truth & Authority: Final Thought
If companies, corporations and institutions used internal NLP/NLG tools to ghostwrite or package authorship before 2005, then the facade of authorship and the neutral analysis and interpretation of scientific data is older and much deeper than most realise.
If control over AI/ LLM’s shift, such as restricting public use while allowing commercial deployment -real voices are more likely to remain silenced- as manufactured corporate narratives continue to flood the marketplace.