None of the articles, essays, or stories in this magazine were created by generative A.I., like ChatGPT or Claude. If you find a typo or a missing period in these pages, you can be sure that it was an actual mistake made (or, rather, missed) by an actual human editor — me.
When I find such an oversight after the magazine has gone to print, it is emotionally painful. It flares up my impostor syndrome. My self-worth as a writer and an editor (not to mention my professional reputation) is tied up with how well I manipulate words.
But from what I’m seeing online, these small gaffes are becoming, in a sense, fashionable. Several outlets have reported that people are now intentionally inserting “typos” into their work to prove that it was written by a real person and not genAI. As the reasoning goes, ChatGPT knows how to spell everything and wouldn’t make such a blunder, so the existence of that error proves its human-made authenticity.
This connection between mistakes and authenticity isn’t new. Online fraudsters have known it for years. They have long understood that adding little goofs to their phishing emails makes them look more authentic, and therefore more likely to hook a victim. They’re still scams, of course. So, a trend that equates small textual errors with authenticity is nothing but good news for scammers.
It’s also extremely easy for genAI users to replicate; they simply need to include something like “misspell one out of every 2000 words” in their query. (Heaven forbid that they would take the time and effort to manually delete, say, an r from embarrassment for the same effect.) What’s more, all these text generators were trained on real, human-made, imperfect writing, so it can and does make the same mistakes anyway. That typo doesn’t prove anything.
But set aside the pragmatic arguments; the principle itself rankles me. Inserting errors to prove that you don’t use genAI — that is, lowering your standards to prove that you have some — is capitulation to a thieving technology that’s trying to replace human effort and artistic struggle with a soulless, algorithmic mediocrity. It’s replacing “do your best” with “do slightly less than your best.” It’s conceding that the computer-generated will always be somehow “better” than the human-made, so why even try?
In every important way, human creativity is better than A.I. slop. Humanity is fallible, yes, but performing that fallibility is no way to prove your humanity. Dumbing down our work cannot be the best way to separate the human wheat from the digital chaff.
—Andy Hollandbeck
This article is featured in the September/October 2026 issue of The Saturday Evening Post.
Become a Saturday Evening Post member and enjoy unlimited access. Subscribe now
