Is That Your Voice—Or Is It AI's? 20 AI Writing Tells Every MBS Author Should Know

20 AI Writing Tells Hiding in Your Spiritual Manuscript

This blog is subject to our Notice to Readers.

AI tools have become part of how many authors write now, and I understand why. You’re working through complex, layered ideas and trying to put something sacred into words. Sometimes you need help getting started, or getting unstuck. It makes sense to turn to AI—Claude, ChatGPT, or Perplexity, for example—to give you a little inspiration.

But as a developmental editor who works specifically in Mind Body Spirit publishing, more and more, I’m seeing manuscripts where the author’s real voice—the one that drew me to their work in the first place—has been replaced by something that sounds like every other writer. AI-generated spiritual writing has a particular flavor. It’s competent, cohesive, and it hits all the expected notes. Yet rather than going deep into a subject, it glides over the surface of things. Readers may not be able to put their finger on why what the author is saying doesn’t seem to land, but they feel it. If you’re self-publishing, using AI won’t matter all that much to your sales at first, because readers will have bought your book before they discover that it’s AI-generated or AI-assisted. But they’ll tell their friends, who’ll tell their friends, and after a while, your book is going to stop selling because, frankly, nobody wants to read same-old, same-old, even if it does have your name on the cover.

But if you’re looking for an agent or a publisher that takes unsolicited queries from authors, I guarantee you they are absolutely going to notice your AI content. And they are not going to take on a book that isn’t at least 80% human-written. And even that number may be changing as we get deeper and deeper into the AI world.

So consider this a loving heads-up. Editors across the industry (myself and top-notch editor Kelly Notaras among them) keep landing on the same handful of patterns. Here’s what to look for—and here’s what not to include in your writing.

1. “It Is X. It Is Y. It Is Z.” Stacking (and the Rule of Three)

This is one of the most reliable AI tells. AI loves to define a concept through a series of short, parallel declarative sentences, each beginning with “It is” or “It’s.” It creates a rhythmic, incantatory effect that sounds spiritual—but it’s actually a structural shortcut. Here’s a real example:

Example: It is a living experience. It is the quiet knowing that the breath we take is shared with the trees, the oceans, and the stars. It is the sense that our joy matters not only to ourselves, but to the whole.

That’s three sentences that essentially do the same thing: restate the concept with a slightly different image. No real development is happening. A human writer building momentum would move the idea forward with each sentence, not just re-dress it.

And notice the adjective “quiet”? “Quiet knowing.” Why quiet? Isn’t knowing always quiet? “Quiet” or other similar modifiers are another tell that AI uses all the time: “quiet + a verb,” “gently + a verb,” “softly + a verb.” As soon as I see that in a manuscript sample, I stop reading. Because personally, I don’t want to read AI-written material.

This is really a specific case of a bigger habit: AI generates tricolons—sets of three—compulsively. Every sentence that could carry three elements will carry three. Human writers mix doubles, triples, and run-ons by instinct; AI tricolons show up so consistently they start to feel structural rather than chosen.

Example: She was hollow, unmoored, and utterly beyond reach.

2. The Soft Pivot Transition (and the Summary Bridge)

Watch for transitional sentences that sound meaningful but don’t actually connect anything. Phrases like:

Example: “When we begin to see life this way, something within us softens.”

Example: “As we open to this truth, we begin to remember.”

Example: “And in that space, we meet ourselves anew.”

These are the connective tissue of AI prose. They’re designed to feel like insight, but they’re essentially filler—a graceful way to move from one idea to the next without actually doing the hard work of showing how the ideas relate. If you read a sentence like this and ask yourself “but what does that actually mean?” and can’t answer it specifically, that’s a flag. It’s probably an AI tell.

A close cousin is the summary bridge: before introducing the next idea, AI restates the previous one in a single sentence—a kind of verbal throat-clearing that reads as padding.

Example: That was the shame she’d been carrying. Now it was time to talk about healing.

Human writers trust their own momentum and move on without announcing the move.

3. The Science-Meets-Mysticism Summary

This one shows up constantly in metaphysical writing, and it almost always reads as AI-generated:

Example: “Modern science is beginning to catch up with what mystics have known for centuries…”

If you’ve seen this sentence—or a close variation of it—in your own draft, please know that it appears in thousands of manuscripts. It’s a default framing that AI reaches for whenever it needs to bridge spiritual and scientific concepts. It’s not wrong, exactly. But it’s not yours, either. And it won’t distinguish your book from anyone else’s.

4. Dramatic Italicized Whispers

AI loves a moment of theatrical italics. A line set apart, formatted for impact, meant to land like a revelation:

Example: It whispers: There is something more. Something only you can bring.

The problem isn’t the device itself—italicized emphasis can be beautiful in the right hands. The problem is that AI deploys it formulaically, as punctuation for a section rather than as a moment of heightened language. If the prose surrounding it isn’t distinctive enough to earn that pause, the whisper falls flat. The dramatic gesture has to be the payoff of something real, not a substitute for it.

5. Staccato Sentences Used as Rhythmic Decoration

Single-word or very short sentences, dropped at regular intervals to manufacture drama:

Example: He walked in. Set down his keys. Wednesday. The world, somehow, still turning.

A few of these land beautifully. Strung together at every scene change, they reveal a stylistic algorithm rather than a writer’s genuine instinct for when silence earns its keep.

6. The Short Declarative Trio

Related to the “It is” stack, but slightly different: a run of punchy short sentences used to create emotional cadence:

Example: “Gratitude makes us patient. It makes us faithful. It roots us in the now.”

This feels rhythmic and resolved—and that’s exactly what makes it suspicious. Genuine insight usually arrives a little more unevenly, a little more surprised by itself. When every three-beat landing sounds this polished, it’s worth asking whether you actually wrote it or whether you accepted a suggestion. A few of these in a human-written manuscript are fine. But don’t overdo it, or your work will start to look like AI.

7. Generic Journaling Prompts

If your manuscript includes reflection exercises or journaling prompts, read them carefully. AI-generated prompts tend to be broad, safe, and interchangeable—they could appear in any MBS book on any topic:

Example: “What am I deeply grateful for—not just today, but in my life?”

Example: “When do I feel most like myself? Most alive?”

Your prompts should be specific to your framework, your language, and your reader’s journey through your particular book. If they could be lifted out and dropped into someone else’s manuscript without anyone noticing—they probably weren’t written with your reader in mind.

8. The Rhetorical Question

AI often asks questions that the reader already knows the answer to. Here are some patterns to watch for:

The “aha” setup: “What if the answer was never outside of you, but had been waiting patiently within all along? The truth is, it was there the whole time, gently waiting for you to remember.”

The stacked triplet: “What does it mean to truly surrender? What does it feel like when the soul finally exhales? And what happens when we stop asking and simply allow? It means releasing the grip of control and trusting the unfolding of what already is.”

The false binary: “Is this a wound, or is it a doorway? In truth, it is both—the wound is the doorway.”

The cosmic invitation: “What if this moment of discomfort is not a setback, but an invitation from the universe itself? When we choose to see it this way, everything shifts.”

The “haven’t we all” bonding move: “Haven’t we all, at some point, felt the quiet ache of disconnection from our true selves? And yet, in that very ache, we find the beginning of our return.”

The rhetorical-into-affirmation pivot: “Could it be that your intuition has been the guide all along? Perhaps it has. Perhaps it always will be.”

What makes these read as AI-generated (versus a human writer using rhetorical questions deliberately) is usually the density and the resolution pattern. A genuine spiritual writer might drop one rhetorical question and let it sit unanswered, or answer it messily and personally. The AI tell is the reflex to immediately smooth it over with a tidy, universal answer, often using the same “not X, but Y” or “what if…instead” scaffolding every time. That last example—“perhaps it has / perhaps it always will be”—is its own sub-tell worth flagging: the reflexive repetition-with-slight-variation instead of just… stopping. A human writer who trusts the reader tends to let the question land and move on; the AI pattern can’t resist closing the loop. AI also tends to cluster these questions as a structural crutch between points rather than as a deliberate escalation—two in a single section is usually where technique tips into tell.

9. “Without X” Phrase Clusters

AI stacks “without” constructions in close proximity, creating the sound of literary restraint while actually demonstrating the opposite:

Example: She entered without fanfare, without ceremony, without the apology she had rehearsed all morning.

Human writers reach for this construction occasionally. AI repeats it until the elegance collapses.

10. “Something” as a Vague Placeholder

Along with “something greater” or “something more” as stand-ins for the Divine, AI leans on “something” generally to gesture at ineffable experience rather than rendering it concretely:

Example: Something had changed in the room. Something between them. Something that didn’t have a name but had a weight.

Human memoirists push through the vagueness toward a specific image or sensation. AI rests in the gesture.

11. Compound Abstract Noun Phrases as Standalone Sentences

These constructions feel like theme statements dressed up as prose:

Example: The particular exhaustion of loving someone who is leaving. The cost of that kind of patience.

They sound weighty but lack the specificity of writing that grew from lived experience. They translate meaning rather than embody it.

12. Overly Neat Simile Stacking

Each individual simile may be technically fine, but the consistency of their construction—clean, contained, never rough-edged or surprising—is a tell:

Example: Like a window left open in winter. Like a song heard through a wall. Like the moment before you remember what you’ve forgotten.

Human writing has idiosyncratic, occasionally slightly clunky comparisons mixed in. AI similes are uniformly well-formed.

13. Engineered Paragraph-Ending “Landing Lines”

AI has mastered the kicker sentence—the resonant line that closes a paragraph with a small flourish:

Example: And maybe that was all any of them had ever needed—permission to begin.

It does this so reliably that the rhythm becomes predictable: build-up, pivot, landing line, white space, repeat. The cadence eventually exposes itself.

14. Domain-Specific Grounding Words on a Schedule

In MBS writing especially, AI will anchor spiritual content with a single “practical” word—practical, tangible, concrete, embodied—at metronomic intervals, as if hedging against its own abstraction:

Example: The practice wasn’t mystical. It was practical. Measurable. A practical tool for a practical kind of peace.

The result is calibrated rather than felt.

15. Clinical Descriptions of Physical or Spiritual Sensation

AI describes embodied experience with symptom-like precision—localized, mild, measurable:

Example: A warmth in the chest region, moderate in intensity, accompanied by a sense of lightness in the upper extremities.

Human memoirists writing about somatic or spiritual experience tend to be messier, more associative, more willing to reach for the strange image. AI is careful not to overclaim.

16. “Gave Herself Permission To” / “Allowed Herself To”

This construction appears with particular frequency in MBS and spiritual memoir writing. It sounds like an insight but functions as a shortcut—a stock phrase that stands in for the harder work of showing what a shift in a character actually looks and feels like:

Example: For the first time in years, she allowed herself to simply rest. She gave herself permission to not have the answer.

This is the plot-level version of a phrase-level habit—the standalone word “permission” shows up constantly as a bonding move too (“this is your permission to…”), so if you’re leaning on the concept of permission at all, look closely at how you’re using it.

17. Abstraction in Place of Specific Detail

Human memoirists anchor scenes with proper nouns: the name of the diner, the year, the song playing, the exact street. AI defaults to generic containers:

Example: They met in a coffee shop in a mid-sized city sometime in her late thirties. She remembers it was raining.

The result floats rather than lands.

18. Interpretive Gloss Inserted After Resonant Moments

AI does not trust a scene to land on its own. After an image or emotional beat, it adds a sentence that names what just happened—closing the loop the reader should have been left to close:

Example: The door clicked shut behind him. In that moment, she finally understood what she had been waiting for all along.

Human writers extend more trust to silence.

19. No Idiosyncratic Authorial Voice

This is the big-picture tell, and it’s structural rather than local—you won’t catch it in any single sentence. The prose is technically competent from first page to last, almost uniformly so, but there’s no grain to it. There’s no verbal habit that belongs distinctly to this person. Human writers have quirks. AI’s voice is smooth in a way that starts to feel like averaging. Look for it across the whole manuscript, not in any one passage.

20. A Few Other Words to Watch For

Over time I’ve built a personal list of words and phrases that show up disproportionately in AI-assisted writing. None of them are wrong on their own—but when several cluster together in a short passage, it’s worth pausing.

Atmosphere & Energy Words

• alchemy / alchemize

• deep / deeply

• gentle / gently

• genuine / genuinely

• liminal

• luminous / luminosity

• quietly / in quiet ways / quietly knowing

• resonance / resonate

• sacred

• tapestry (especially “tapestry of life / connection / meaning”)

• threshold

• unfurl

• weave / woven / interwoven

Journey & Process Words

• attunement

• awakening

• embodiment / embody

• excavate / excavation (of the self, emotions, etc.)

• integration

• journey (used as a noun where “path” or a more specific word would serve better)

• remembering (as in “remembering who you truly are”)

• unfolding

Connector & Transition Phrases

• “In this space of…”

• “Holding space for…”

• “At the intersection of…”

• “It’s not X, it’s Y” constructions (e.g., “It’s not about doing more, it’s about being more.”)

• “Not a X, but a Y” (e.g., “not a breakdown, but a breakthrough”)

Validating & Bonding Phrases

• “you are not alone in this”

• “this is your invitation to…”

• “permission to…”

• “deeply human”

• “profound truth” / “profound shift”

Vague-but-Weighty Nouns

• container (as in “holding a container for healing”)

• portal

• medicine (metaphorical, e.g., “your story is medicine”)

• wisdom, overused as a standalone noun (“your inner wisdom,” “ancient wisdom”)

• sacred exchange / sacred recognition / sacred [blank]

• “something greater” or “something more”

Closing Affirmation Constructions

• “I am grateful for…”

• “I walk in…”

• “I choose…”

These words and phrases often appear stacked in twos or threes at the end of a chapter or post, functioning as a ritual close rather than an earned conclusion.

The Dreaded Em Dash

Em dashes have become one of the clearest AI tells in writing today—and the irony is that human writers have used them forever, effectively, and sparingly, to add a beat of emphasis or a quick aside. The problem is frequency and rhythm: AI uses the em dash as a default connector, stringing clauses together—one after another—in a way that starts to feel less like considered punctuation and more like a stylistic tic. A human writer might use one or two per page, deliberately, because a comma or period would feel flatter. AI writing can use four or five in a single paragraph, turning what should be a moment of emphasis into background noise. The fix isn’t to ban the em dash—it’s a useful piece of punctuation—but to use it the way a careful human writer always has: rarely, and only when nothing else will do the job. This is a sad story for writers, because the em dash is a beautiful writer’s tool, and now (at least for the time being), we writers are going to have to let AI have the em dash. For now, periods, semicolons, and brackets will need to be our tools for asides and exceptions. It won’t be forever, just for now. The em dash shall rise again.

So What Do You Do With This?

First: don’t panic, and don’t delete everything. AI can be a genuinely useful drafting tool. The goal isn’t to scrub it out, it’s to go back through and ask yourself, honestly: is this what I actually think? Is this really how I say things? Does this sentence sound like me on a good day, or does it sound like a reasonable approximation of an MBS author?

Your real voice:

• is specific

• has a relationship with the metaphors you’re using

• makes unexpected connections

• hesitates in places and rushes in others

• occasionally says something slightly weird or overly personal or not quite resolved (and those are often the best parts)

When I’m reading a manuscript and I feel something shift—when a sentence surprises me, when I catch myself nodding, not because I’ve heard it before but because it’s true in a way I hadn’t considered—that’s you. That’s the voice you want to protect.

Your book deserves your voice. Your readers are reading your book because they want to hear from you—not from Claude, or ChatGPT, or Perplexity. You. So use AI as a scaffold if you need to. Just make sure that when the building is done, you’ve replaced the scaffold with your own walls.

And by the way, I wrote this blog with the assistance of Claude. I gave “him” a page filled with detailed notes on what I wanted to say, together with several partially completed and fully intact paragraphs, and then asked “him” to flag “his” own AI tells. Claude assembled my notes, expanded on them, rearranged them, and generated a first draft. Then I went through it and rewrote sections in my own voice. So Claude and I wrote this blog together—about 50/50. Like I said, AI is a wonderful tool. Claude saved me time.

Can you tell the difference between what Claude wrote and what I wrote? Good. Because as a professional writer, you’ll need to get good at distinguishing AI writing from human writing. Maybe not today and maybe not tomorrow, but soon, human writers are going to put their proverbial foot down and lobby bookstores (online and brick-and-mortar) to stop selling AI-written books. (Some already don’t, as a matter of policy.) And one day, readers are going to stop buying AI-written material. See an AI-written book you like online? Why spend money on it, when you can just give your own AI app a prompt on the given topic and let it write you a book on the same subject? Easier, cheaper… and forgettable. Don’t let that be your book.

That’s the world we’re heading into, and it’s not coming someday—it’s already knocking. The writers who survive it will be the ones whose voice can’t be prompted into existence by someone else. So find yours. Protect it. And don’t let a machine, however helpful, write the parts only you can write. Readers want to hear from you. The world is waiting for what only you can say. Go say it.

Michael Ireland

michaelirelandeditor@gmail.com

And in the interest of full disclosure, I wrote this blog about how to identify AI-generated material with the assistance of AI. Thanks, Claude! It’s a strange world we live in. 

Thanks too to editor Kelly Notaras for being my editorial guru.

---

This blog is subject to our Notices to Readers section.

Please click to review the Blog Notice to Readers.

---

For more information and helpful tips on editors and publishers, check out my other blogs. Questions? Reach out at michaelireland@shaw.ca.

---

P.S. Want to read my blogs on mystical, magical, and metaphysical topics?

Visit me on Substack: Substack.com/@michael852085