Why the "It's not X, it's Y" AI tic backfires on its writers
by Ellsworth Toohey · Boing BoingTry to keep the idea of a white bear out of your head.
Daniel Wegner's 1987 experiment is now showing up in a critique of one of the most recognizable AI writing tics, the LinkedIn-saturated "It's not X, it's Y," by University of Guelph linguistics scholar Joshua Gonzales, writing for The Conversation.
Cognitive psychologists have tested this since at least 2003. Readers process the negated noun first, with the intended alternative arriving only after extra processing time. A 2004 study added a wrinkle. If readers can supply an obvious substitute ("not guilty" triggers "innocent," "not cold" triggers "warm"), the brain swaps in the substitute. Without one, the negation tag drops off in over a third of cases, and people remember the affirmed version.
Which is bad news for "This isn't marketing, it's a movement." Marketing has no clean opposite. Readers store "marketing," Gonzales writes, "with a sticky note that's already peeling off." The reframe that was supposed to make you forget marketing makes you remember it.
A 2024 study on AI-assisted writing found that texts produced with AI help were about 10% more similar to each other than human-written ones. Once a phrase like "It's not X, it's Y" spreads across LinkedIn, it stops being a quirk and becomes the basic shape of how ideas get introduced.
Gonzales's fix: "Say what it is. Say what you built, what you believe, what you offer."
Previously: