Everyone is Secretly Using an AI Ghostwriter

The following introduction was written by GPT-5.5, answering a fairly detailed prompt from myself. I ask that you read it through narrowed eyes, the way you would investigate a particularly well-spoken murder suspect whose attempts to act like everything is normal are a little too good.

#Introduction

"Everyone is writing with a ghost now. Not the old kind—the silent editor, the borrowed cadence, the mentor whose sentences we unconsciously mimic—but a machine that finishes our thoughts before we have fully formed them. Emails, essays, captions, articles, pitches, posts: more and more of the language we encounter has passed through AI’s hands, if it was not produced by them outright. The result is not merely that people are outsourcing their writing. It is that the shared atmosphere of writing is changing. We are all breathing in the same polished, frictionless, vaguely competent air.

Even those who refuse to use AI are not safely outside its influence. Style is contagious. We write, in large part, from the composite voice of everything we have read: the books that trained our ear, the essays that sharpened our rhythm, the strangers online whose phrases lodged themselves in memory. You are what you eat, and your writing is what you read. So if AI becomes the most common author in our daily lives—if its cadence, structure, and bland confidence become the background music of modern language—then resisting the tool will not be enough. Unless we are careful about what we consume, we may find ourselves writing like AI without ever asking it to write for us.

Large Language Model With No Soul (AKA GPT-5.5)

#(Actual) Introduction

Unfortunately, you might not have known that was AI if I hadn’t told you. You might simply have thought my writing ability to be that bold and punchy. The valiant short sentences might have sold you; the mock courage rendered platitudinous by excessive hedging might’ve pulled you in. Indeed, Large Language Models seem to not-this-but-that (to coin a verb) with such reckless abandon that it portrays its prompter as the sweetest, most reasonable gentleman ever to put pen to paper––or finger to key.

The advent of cheap, often free, Large Language Models has turned everyone into a writer. It no longer matters if you spent your childhood cultivating a habit of creative writing, with dreams of one day using your skills to bless the general public with stories that plumb the depths of the human spirit and elevate divine truth, goodness, and beauty. Now, you can simply type the following prompt into ChatGPT…

sooper good story that awesomely rittun only hundrad wurds max the sort that win prize comptishun

…and its gears will churn for nine seconds before spitting out…

Every morning, Mara wound the town clock, though no one paid her.

“Time moves itself,” they said.

On her ninetieth birthday, she stopped.

At noon, bread refused to rise, trains forgot to leave, and children hung mid-laugh in the square. Only Mara could move. She climbed the tower slowly, each step a year.

At the top, beside the gears, she found a small note in her mother’s hand:

Keep going. They notice only when love stops.

Mara wound the clock.

Below, the town applauded, thinking it thunder.

Whether this is a good story or not (let alone that coveted descriptor, ‘sooper good’) is up for debate; yet it could justly be labelled ‘creative’ by any judge unaware of its origin. That is what concerns me.

In this article, I aim to demonstrate why using AI to write for you is evil. Not just a bad idea, but evil. More than this, I hope to show that even if you do not use it to write for you, if you merely read the outputs of Large Language Models on a regular basis it can have a destructive effect on your creativity in every area of life. More positively, I will demonstrate what you can do about the problem of AI in writing, and propose a long-term global solution that is currently beyond my capacity to implement.

#First, A Confession

In writing an article like this, I would do well to first confess where I myself have sinned in my use of AI. Two areas come to mind.

The first area relates to the AI-generated artwork which brazenly adorns many of the earlier posts on this archive. This, I believe, was a form of stealing from the real artists and photographers from whose work the Large Language Models have often drawn without permission. This was sinful; I ought not to have used it.

Second, I also used AI as an editor and advisor in many of the earliest posts on this blog. Inasmuch as I was assisted in the technical and mechanical editing surrounding grammar, punctuation, and syntax, this was not sinful. It was no doubt unwise, however, given how difficult it is to distinguish between the purely Microsoft-Word-like spellchecking and the morally-informed advice and rephrasing which Large Language Models often aspire to provide. To the extent to which I sought out the phrasing, structuring, and moral advice of AI, I confess that this was sinful; it was lying. That extent varied greatly—from minor grammatical checks to rephrasing of entire paragraphs. It would unfortunately be impossible to identify and confess each specific example, since I would not be able to put a fine line between my own original writing and the adjectives, phrases, names, etc., introduced by AI. I simply do not remember.

My hope is that there is sufficient evidence of my fascination and competency in writing before November 2022 to prove that my above use of AI was at worst supplementary and does not represent my ability in this field.

(Why November 2022? Because…

“LLMs attracted substantial public attention after OpenAI’s ChatGPT was made publicly available in November 2022.”¹

I, for one, remember being introduced to ChatGPT in December 2022 by my brother.)

Moving forward, I hope you will forgive me for these sins and accept the following article as evidence that, praise the Lord, I now hate them.

#Why Using AI to Write For You is Evil

“In the beginning was the Word, and the Word was with God, and the Word was God” (John 1:1 LSB).

#Sins of Words

Christians worship God—the God who is the Word. To put it another way: Christians worship the Word. We are people of the Logos. Any sin related to the use of words, therefore, is not to be treated lightly because it stabs at the heart of the Christian faith itself.

Scripture is clear about this:

“But I tell you that every careless word that people speak, they shall give an accounting for it in the day of judgment. For by your words you will be justified, and by your words you will be condemned” (Matthew 12:36-37 LSB).

“If anyone does not stumble in what he says, he is a perfect man, able to bridle the entire body as well” (James 3:2 LSB).

Sins related to words, therefore, are worth attention; they are worth a fuss.

#Creation is an Act of Words

Now, when the Word was with God in the beginning, what was He doing? He was creating (Jhn 1:3, Gen 1:3). The universe, at its root, is not made of stuff; it’s made of words. The trees and bricks and motherboards and lakes you see around you are fundamentally comprised of the spoken words of the Word.

Then God made man in his image. Unlike the fish, the beasts, the trees, and the light, man uniquely had the divine gift of creation and dominion. It was he who would name the animals, he who would take dominion, he who would fill the earth. The image of God, the Imago Dei, means you, like your Creator, are a creator. Not a capital ‘C’ Creator; your ‘c’ is lowercase—sometimes very lowercase, if you know what I mean.

Be that as it may, any creation of yours is made of your words—at least inasmuch as you created it. The truth of this is plain when we consider a novel: the ‘existence’ of the fictional world hangs upon the words that emanated from you. No words means no novel means no fictional world.

But what about a bridge you designed? In what sense does it hang upon your words? It does so in the sense that the logical (think: logos, word) processes that went on in your head produced that plans that oversaw the bridge’s construction. The physical reality of the bridge still depends upon God’s physical world—which depends upon His words, of course. But there is a real dimension of the bridge’s existence that depends upon yours as one of God’s creatures endowed with creative ability.

#Creation and Lying

Suppose, however, that once the bridge had been laboriously constructed (several honest workers getting injured in the process), you proceeded to waltz around to all and sundry, vaunting the phrase, “I built the bridge! I built the bridge!” It clothed your LinkedIn page; it adorned your CV; it sat like a crown of glory upon the head of any conversation other men were unfortunate enough to engage you in.

Now, unless you really did take part in the physical labour, there is a good word to describe your behaviour: you lied.

Your lie, however, is worse than a simple “No I didn’t eat the last biscuit.” It goes much deeper. You, as a divine image-bearer—a creator like God, are misrepresenting God Himself in your untruthfulness. You are saying it would have been okay for God to lie about the source of all creation. Perhaps God really only designed the world, but didn’t bring it into existence; perhaps all the stuff was lying around already and He merely fashioned it; perhaps He left the actual hard work to the angels.

#By the Same Token…

I believe the analogy is clear. We are not really talking about bridge building. My aim is to utterly condemn those who pass off AI-generated text as their own. Those who do this unrepentantly are liars.

“But for the cowardly and unbelieving and abominable and murderers and sexually immoral persons and sorcerers and idolaters and all liars, their part will be in the lake that burns with fire and brimstone, which is the second death” (Revelation 21:8 LSB, emphasis added).

If you are in this position, confess your sins to God and (inasmuch as you can) the people you have wronged by lying to them. This will generally mean that your repentance ought to be as public as your sin. If your Facebook posts were AI-generated, confess this as sin on Facebook. If your sermons were written by AI, confess this in a sermon. And etcetera.

If you are unsure whether your use of AI is sinful, ask the Lord to reveal to you your sin. A good rule of thumb is: would I feel shame if my exact use of AI were abundantly clear to all?

#Who Are You Writing For?

Having established that using AI to write for you is a form of lying we can move on to a demonstration of why it is not just wrong, but stupid—foolish.

I will draw from my own experience and temptations here. The times I have been tempted to use AI to write are the times I have been seeking the acclaim of men more than the acclaim of God. I have been more desirous of the short-term outcome (immediate approval from my readers) than the long-term outcome (utilising or developing my own writing skill to the glory of God). Often, I am in the process of writing out an argument that is difficult to make, difficult to pull off, and being unable to tie the threads together in a satisfactory manner, I am tempted to reach for forbidden fruit—to ask the Large Language Model to do it for me.

But where is the gain in such cases? What does it achieve? As author N.D. Wilson puts it, it is like having a robot bench 150 kilograms then boasting as if you did it. Do you have no shame? Or are you just dumb?

Some Reformed writer may attempt to justify his deception by saying, “For the sake of the public good I am releasing this AI-written article—because no one else has published a defence of hyper-sub-infralapsarianism with quite the same scrupulous attention to detail. It has to be done.” This, however, is no more than a lack of faith in God’s sovereignty; not only are you in sinful despair over the state of the world as God ordained it (it being notably and deplorably empty of all sound defences of hyper-sub-infralapsarianism), you are then presuming to help Him out a bit by cutting corners—breaking his law. Don’t.

The problem is thus reduced to one question: who are you writing for? Or, to be grammatically fastidious: for whom are you writing?

“For am I now seeking the favor of men, or of God? Or am I striving to please men?” (Galatians 1:10 LSB).

If you write for the Lord, not men, you may find that your prose leans closer in style to the book of Galatians than the effeminate posturing and hedging that is characteristic of AI. There will be unvarnished streaks of courage in your style. It will smell like sweat. It will breathe because it has a soul. When you knock on it, you will find that it answers because someone is home. In short, you will write like a human being who has a soul—and this will glorify God.

#Objections

“But the ideas were mine; AI just wrote the words.”

Do your readers know this? If not, it is rank deception.

Even if they do know, does the benefit they gain from imbibing your ideas outweigh the harm their creativity suffers from reading AI-generated text? (More on this later.)

“I wrote the article; AI just edited it.”

Let’s be real about this. It probably didn’t “just edit it.” The moment AI is offered the position of editor it hands itself a promotion, suggesting forty-three phrases you should reword, twenty-five metaphors that are ‘unnecessary’, and seventeen arguments that were a little too harsh (assuring you, of course, that it doesn’t want to flatten your style, just ‘direct’ it).

“AI knows my writing style; I’ve set it up to write just like I would if I had the time.”

This really gets to the nub of the issue—and perhaps I ought to be emphasising it more in this article.

When you write—as in, when your brain spends the effort to produce the words—the written creation is necessarily derived from God’s image, and thus bears the marks of God’s image. It exists on a spectrum of truth, goodness, and beauty because a moral agent created it who is capable of grasping these transcendental qualities. I do not say that these qualities are necessarily exalted in a man’s writings; what I mean is that they have the capacity to be exalted or abased, glorified or spurned, because the man can be righteous or sinful.

When AI generates text, however, it is devoid of the Imago Dei. God is thus nowhere present in the words themselves—neither in glorification of Him, nor in dishonour of Him. He may be dishonoured in the fact of the text having been generated, or the objective content of the text, but the words themselves are soulless and cannot be adjudicated as arising from a moral agent.

Even if AI does write “in your style,” it lacks the image of God by definition. The writing did not arise from a mind and that matters.

Please dwell on this idea more for I have not even begun to plumb its depths.

#The Destructive Effect of Reading AI-Generated Writing

Many well-intentioned writers and authors (myself included) run the risk of finding themselves writing in a style eerily similar to AI and wonder how it happened. It is made all the worse by the fact that if you write in the style of AI, then you have been trained to think in the style of AI—because where other than your thoughts did your writing come from?

The explanation for this phenomenon is simple: just as you are what you eat, you also write what you read. Your influences really do have an influence on you—and AI can be one of those influences just as easily as Lewis or Tolkien.

When I write, for example—and particularly when I am at a loss for words—I frequently ask myself questions like, “How would C.S. Lewis put this?” or, “What metaphor would P.G. Wodehouse use here?” However, I remember the shock horror that overtook me when I caught myself thinking, “Would ChatGPT approve of the harshness of my polemical style here?” It acted like a bucket of ice poured over my hot shower (okay, sure: Wodehouse would do better, but it serves the purpose for now).

The more writers read AI-generated text, the more templated and uncreative their writing becomes. You have no more control over this than you do over where a donut goes when you eat it.

Of course, if you only read AI text occasionally—skim-reading when you do, and squinting your eyes in the manner I described at the beginning of this article—it is possible to live in an AI-saturated world without harm to your creativity. You can retain the soul of your writing so long as you are intentional about doing so.

But unless people are honest about their AI usage, how can we know when we are reading it?

#An Objection

“AI-generated text is still based on its large database of human text, so it’s basically just spitting out reorganised human text.”

And I suppose chicken nuggets are good for you because they’re basically just chicken.

#A Proposed Solution

Be sure your sins will find you out. You cannot, and will not, manage to hide forever the origin of your style. If not now, then later there will be a way to know who is and is not outsourcing their literary thinking to a bot. If I had the time and resources myself, this is how I would do it:

1. Like the blue tick on Facebook and X, authors and artists would be issued metaphorical blue ticks to verify their competence.
2. (It would be nice if the metaphorical blue ticks took their literal form as gold-plated, handwritten certificates written with fountain pens, set in a wooden frames that smell of garage.)
3. To obtain your metaphorical blue tick, you must demonstrate your skills (whether that be writing, painting, music, or any other art) in-person to a silent, live audience, doing so under conditions similar to jury duty (especially if the project takes multiple days).
4. At the end, your completed project is not marked, nor evaluated; it is simply acknowledged as existing by the live audience, who each sign off as having seen you create it.
5. The public should then have easy access to read/view/listen to at least a portion of this completed work as proof of your competency.

The most obvious objection to this idea is the fact that it would take many man-hours to pull off with almost no obvious financial recompense. I acknowledge that this objection is thoroughly sound, adding only this question: “You got a better idea?” Perhaps a society’s willingness to expend its time and money in protection of art is a good measure of that society’s health. If our society is not prepared to take drastic measures to protect the sanctity of good art, it is a society that does not deserve that art.

Those who want to use AI to detect AI content are only compounding the problem. This will lead only to an arms race, and as such, one could never be quite sure if those in charge of the AI detectors or those producing the AI content have the latest advances in the technology. At the end of the day, we don’t know for sure whether China or the US would win a war until/unless that war is waged. There are too many unknown variables.

The battle for art and creativity is best fought face-to-face, using the senses God gave you.

#Possible Ethical Uses of AI While Writing

Here’s one:

“Perform the following task and respond in the most succinct, no-fluff manner, like the robot machine you are. Read the following article and flag every place where I have accidentally used a lowercase h “him” when referring to God. Do not include any extra text whatsoever.”

The above request is representative of all manner of mechanical tasks that AI could perform that a simple spellchecker couldn’t—like incorrectly used homophones.

#Conclusion

As a Christian, it essential that you do not have an evil eye of suspicion against your brother. I have fallen into the trap of accusing others of using AI in writing when, though I may have been right, I had no way to prove my accusation. Accusations that do not have two or three witnesses are sinful. Do not make them.

Recognise, then, that the only way of knowing if an article or book has been written entirely by a human is if you know and trust the human himself (or at least, know of and trust).

And until a solution such as I have suggested above has been implemented, please join me in predominantly reading the writings of men from before November 2022.

#Notes

1. Jan Clusmann et al., “The Future Landscape of Large Language Models in Medicine,” Communications Medicine 3, article 141 (2023), abstract, https://doi.org/10.1038/s43856-023-00370-1.