Regarding AI, Part Two
by Evan Braun
The Mechanical Tells
When AI first came on the scene a few years ago, there was a lot of talk of who was going to be negatively impacted. I found many of those predictions to be alarmist. An obvious example was graphic designers. And of course there were immediate concerns about writers, particularly those in journalism, corporate communications, and advertising.
Yes, many people did point out that editors like me might struggle to find work. In reality, though, my career has grown steadily over the last few years, so the worst-case scenarios, at least in my sphere of influence, never materialized.
That said, something in my professional world did shift in the fall of 2025.
There was one particular manuscript that came through Word Alive Press that caught my attention. At first, it was for all the right reasons—it covered a fascinating subject, doing so very thoroughly, and with evocative language. In short, I was excited to work on it!
But as I got underway, I noticed something amiss. Everything from the book’s structure to its choice of vocabulary was extremely… repetitive and predictable. It occurred to me that no author with the innate skill to have written this book would also have made the mistake of falling into such obvious rhetorical patterns.
By the time I hit the one-third point of this manuscript, I found myself barely able to focus on the underlying ideas anymore. With each passing page, I found the book to be borderline unreadable.
And that’s also right around the time when I diagnosed the problem. After some independent research and consultation with other editors and writers, I felt confident in determining that this author had largely constructed the book through the use of AI. It explained so much about why the prose just didn’t feel right.
I didn’t level any accusation against this author, and I felt no righteous indignation over it. If the author did indeed employ AI, it didn’t trigger any judgement. As far as I’m aware, the book proceeded to be published as per normal.
That said, the very next book I edited, from a completely different author on a completely different subject, mysteriously had a very similar constellation of symptoms.
The following month, it happened again.
Just like that, it seemed to me that the AI conundrum had at last come to my doorstep. I suppose it was just a matter of time.
As I said, I’m not against AI use, but I can’t help but feel these three books, not to mention the many others that came after them, don’t live up to their potential.
So what are the proverbial fingerprints of AI that I started to pick up on? I want to spend some time getting into the specifics, to pull back the curtain on what we’re noticing, editorially, when AI-assisted books come through the queue here at Word Alive Press.
There are two broad categories of AI clues: mechanical and stylistic. In today’s post, we’ll focus on the mechanical ones, which are easier to spot.
Note that rarely would all of the following cues be apparent in one manuscript, even if it were heavily AI-assisted. These techniques are often used independently of AI. But once you start to see them used frequently in combination with each other, they become hard to miss.
1. Ubiquitous lists of three.
Almost constantly, AI text defaults to common rhetorical techniques, primary among them the list of three, which often shows up more than once in a single sentence. For example: “True leadership, biblical stewardship, and godly character require vision, resilience, and compassion.”
2. Antithesis on repeat.
This is one of the clearest tells. AI is known to offer an almost endless series of comparisons, and the cheapest and easiest form of comparison is antithesis, which is the act of defining something by describing its opposite. For example: “Success isn’t about working harder; it’s about working smarter.” In AI manuscripts, you’ll often see antithesis stacked upon antithesis.
3. Overuse of similes.
This is connected to the previous one, in that it also revolves around the practice of reverting to comparison. For example: “His determination spread through the team like a wildfire racing across dry grass, and the grass was as brittle as delicate blown glass.” A simile in itself shouldn’t trigger suspicion of AI use, but an overreliance on this device could be a sign. That example sentence I just used employs two of them, which I would never advise. I recently opened a manuscript in which every other sentence included a simile, which would be unlikely for a human working on their own. As is so often true, less is more.
4. Uniform paragraph length.
In AI-assisted manuscripts, there is often an odd phenomenon where the paragraphs mysteriously settle into the same visual footprint, regardless of the underlying subject being written about. A human author will produce a more organic mixture, whereas an AI book might contain paragraph after paragraph of almost identical length.
5. Symmetrical sentence construction.
This becomes noticeable when clauses in a sentence mirror each other almost too neatly, causing the prose to feel architecturally perfect rather than natural. For example: “She looked for certainty in the future while he found peace in the present.” Done well, and in isolation, symmetrical construction pulls off a nice authorial trick. Overdone and it becomes somewhat empty.
6. Excessive signposting.
With this technique, every transition is announced. For example: “First, we’ll turn our attention to…” or “Next up…” or “With that point established, let’s now explore…” It’s almost as though the reader is being constantly escorted from one room to another, being taken on a carefully orchestrated tour.
7. Balanced cadence.
When handled formulaically, sentences may alternate long-short-long-short with suspicious consistency, as though the prose is being rhythmically arranged. For example: “Some challenges demand patience. Others require courage. The wisest moments call for both.”
8. Reliance on rhetorical questions.
AI prose often uses questions as a structural device to keep the reader moving. Asking rhetorical questions, of course, is a perfectly valid technique. But if you encounter one every few paragraphs, it no longer feels natural.
Again, it’s important to remember that all these techniques are human inventions, and all authors use them to one degree or another, usually without even thinking about it. None of them are exclusive to AI, not by a long shot, and it would be disingenuous to imply otherwise.
But when used in combination with great frequency, sustained almost uniformly across an entire manuscript, these techniques tend to be suggestive of heavy AI influence.
In the next post, I’ll turn my attention to the other category of AI clues: the stylistic ones.
* * *
Did you enjoy this post? Discover more helpful advice from Evan Braun here or read the entire series on AI as posts become available.
About this Contributor:
Evan Braun is a full-time author and editor. He has authored three novels, the first of which, The Book of Creation, was shortlisted in two categories at the 2012 Word Awards. He has released two sequels, The City of Darkness (2013) and The Law of Radiance (2015), completing the series. Braun is an experienced professional editor, and has worked with Word Alive Press authors since 2006. He is also a regular contributor at The Fictorians, a popular writing blog.