March 7, 2026
Witch hunts and em dashes
We're Training Students to Write Worse to Prove They're Not Robots
Kids “write less good” to dodge robot alarms — and the comments are on fire
TLDR: AI detectors are pushing students to simplify their writing — and some even use AI to prove they’re human. Commenters blast the “robot police,” calling it a witch hunt that punishes good writing, while others argue schools should ditch detectors and rethink how they assess real learning.
Students are reportedly dumbing down their essays to avoid being labeled bots — yes, really. One kid’s paper got flagged “18% AI” for using the word “devoid.” Swap it for “without,” and poof, it’s clean. Writing instructor Dadland Maye says this paranoia is so bad that some students started using AI tools just to check if their human writing gets flagged. That’s the Cobra Effect: a tool meant to stop something ends up making more of it.
The comments section came in swinging. One camp says this isn’t new — schools have long rewarded formula over flair. As Paracompact grumbled, standardized tests loved cookie-cutter writing. Others went full scorched-earth: if software’s grading kids, why have school at all? themafia’s “keep them at home” rant drew applause and side-eye.
But the most viral mood was pure witch hunt energy. Someone1234 confessed they now write “less good” on purpose — even online — to avoid being accused of being a bot. Em dashes? Public enemy number one. Big words? Suspicious. Meanwhile, noemit pushed a calmer fix: redesign assessment around how people actually learn, not how to catch cheaters. j45 kept it wholesome with “read more, write better,” plus a wink that those reading muscles will help with AI too. The thread split between “trash the detectors” and “rebuild the whole grading system,” with a side of memes about punctuation on trial and vocabulary getting canceled for sounding too smart.
Key Points
- •A school-installed AI checker flagged a student’s essay as 18% AI-written due to vocabulary like “devoid,” which dropped to 0% when changed to “without.”
- •Writing instructor Dadland Maye reports AI detection policies are pushing students to alter style and even begin using AI defensively.
- •Students are using generative AI to test whether their own human-written work will be flagged by detectors.
- •The article likens AI detection’s unintended effects to the “Cobra Effect,” where incentives worsen the original problem.
- •A high-performing student used Google Gemini to learn what triggers instructor suspicion and AI detection, including sentence patterns and stylistic confidence.