BertaCollins
New member
- Joined
- Mar 9, 2026
- Messages
- 5
My friend Maya (not her real name but she'd kill me if I used it) got THE EMAIL last week. You know the one:
The Backstory:
Maya is not an AI user. Maya is a non-native English speaker who learned English from academic journals. She writes like this:
Turnitin's AI detector said: 72% AI-generated.
The math ain't mathing, people.
The Meeting:
Maya showed up with:
Verdict: "Inconclusive. We'll allow the grade to stand."
"Inconclusive." Not "you're innocent." Just... "we can't prove you're guilty enough to punish you."
The Lesson:
AI detectors are LIARS. They punish:
Screenshots of everything. Every draft, every version, every late-night cry session captured for posterity.
Google Docs version history on lock. (Pro tip: name your versions so it's not just random timestamps.)
Personal voice injections. I now add at least one slightly unhinged metaphor per paper. ("The data flowed like my tears during finals week.") Can't flag that as AI. 
Saving "process evidence." Emails to professors, discussion posts from the same week, anything that shows my writing style over time.
The Question: If you've been falsely flagged, what worked in your defense meeting? What didn't? I'm building a case file for when it inevitably happens to me.
Cue the panic sweats.
The Backstory:
Maya is not an AI user. Maya is a non-native English speaker who learned English from academic journals. She writes like this:
That's just... how she talks. She's been writing like that since high school. It's her voice.
Turnitin's AI detector said: 72% AI-generated.
The math ain't mathing, people.
The Meeting:
Maya showed up with:
- Google Docs version history (17 days of edits, all at reasonable human hours)
- Printed drafts with handwritten notes (old school but effective)
- Her phone (to prove she doesn't even have ChatGPT installed)
- Me (as moral support, silently judging the committee)
Verdict: "Inconclusive. We'll allow the grade to stand."
"Inconclusive." Not "you're innocent." Just... "we can't prove you're guilty enough to punish you."
The Lesson:
AI detectors are LIARS. They punish:
- Non-native speakers
- Neurodivergent writers
- Anyone who writes formally
- People who had good teachers in high school
Screenshots of everything. Every draft, every version, every late-night cry session captured for posterity. Google Docs version history on lock. (Pro tip: name your versions so it's not just random timestamps.) Personal voice injections. I now add at least one slightly unhinged metaphor per paper. ("The data flowed like my tears during finals week.") Can't flag that as AI. Saving "process evidence." Emails to professors, discussion posts from the same week, anything that shows my writing style over time.The Question: If you've been falsely flagged, what worked in your defense meeting? What didn't? I'm building a case file for when it inevitably happens to me.