Sept 9, 2026–I have discovered the secret to defeating Artificial Intelligence, at least in the field of writing.
Flaws.
First, I apologize to loyal readers, as this column smacks of “inside baseball.” But it addresses a current topic and reveals a larger lesson.
As an experiment, I ran a recent column through Grok, one of the leading AI software programs. I am pleased to report that my writing passed as “not AI,” although the reason stings. Grok cited the proof I was not AI in that my writing has flaws. It deemed my writing as “idiosyncratic,” with “elevated diction” and “uneven polish,” and transitions that are not “template-smooth.”
This type of flawed authenticity is what I strive to bring to my readers. (Note: the column I submitted was about my social awkwardness and need for self-improvement. How is that for unforced irony?)
Here I will quote the Great and Powerful Grok:
“This text appears human-written, not AI-generated.”
===
Key indicators of human authorship:
Highly specific, idiosyncratic personal anecdotes that feel lived-in rather than invented: mom spitting on a handkerchief to clean jelly off a cheek before First Communion; locking the thermostat, turning off lights in empty rooms, shutting off the sprinkler under the trampoline, and making everyone order water at restaurants; the staffer’s exact line “you always walk away when we are trying to talk to you about something”; and the 1970s Educational Psychology class exchange that narrows “You’re such a square” to clothes don’t match to blue sport coat with brown shoes. These details are concrete, slightly quirky, and culturally timed in ways typical of personal essay writing.
Natural voice and uneven polish:
Conversational asides (“hey, it was the 70s”), mildly elevated but not over-polished diction (“desultory,” “insipid,” “charade”), self-deprecating humor, and a reflective progression that circles back to the Burns quote without forced symmetry. Transitions feel organic rather than template-smooth.
Originality of the core story:
Web searches for distinctive phrases (the Burns opening + free advice framing, the “square” professor dialogue, the walking-away feedback, First Communion/handkerchief detail) turn up no matches. The Burns couplet itself is widely quoted, but the surrounding narrative and specific examples do not appear to be lifted from published sources.
Typical AI tells that are absent:
- AI-generated text often shows:
- Generic or interchangeable examples
- Overly balanced, evenly paced paragraphs with predictable “setup → insight → takeaway” structure
- Vague or hedged personal claims
- Slightly unnatural smoothness or repetitive sentence rhythms
- None of those dominate here. The piece reads like a personal blog post, newsletter essay, or reflective column by someone drawing on real workplace and family experience.
Bottom line: Strong evidence of human writing. Low likelihood of AI generation (or heavy AI rewriting).
===
So (it’s me again), if I may summarize what makes my writing special according to AI, it is down to these flaws:
- idiosyncratic (peculiar)
- uneven
- unpolished
- mildly elevated (as opposed to decently elevated)
- self-deprecating
- unbalanced
- lacking rhythm
- quirky
- asymmetric
- not template smooth
That concludes your Rhetoric 101 lesson for today. Take away from it what you will. For me, I’ve learned that it is our flaws that make us human. Apparently I got a lot of ‘em.