My design thesis got a review flag last week and I need to understand what’s actually happening here before I respond to my advisor.
For context: my thesis written sections are mine - I use ChatGPT for research summaries and to help outline arguments but I write the actual prose myself. The flag was on a 2,000 word methods section I wrote from scratch in one sitting.
Questions I’m trying to answer:
Does Turnitin actually detect ChatGPT-generated text or is it detecting writing patterns that happen to correlate with AI usage (like formal academic register, certain transitions, structured argumentation)?
Can original writing that’s heavily informed by AI-assisted research trigger detection?
What’s the actual process when a flag comes up - is it automatic review or does a human make the final call?
Has anyone been through a similar situation at a design school specifically?
I’m not trying to get away with anything, I genuinely believe my work is original. But I need to understand the system I’m dealing with before I have the conversation with my advisor.
The detection vs pattern-matching distinction you raised is the right question. Most AI detection tools (including the one layered into Turnitin) are probabilistic, not deterministic. A score is not a verdict - it’s a risk flag for human review.
Design schools vary a lot on process. Some have it go straight to academic integrity review, others have the instructor review first. Ask your advisor directly what the next step is. Being proactive about the conversation is better than waiting.
@voidvibes92 heavily structured, formal academic writing can score higher on detection even when it’s original - the model was trained on a lot of academic-style AI output. Tonal similarity to training data gets flagged regardless of actual origin.
Keep records of your drafts - timestamps, version history, any notes or research you made. If it goes to formal review, showing your process is the strongest evidence. Digital breadcrumbs matter here.
The answer to “does it actually work” is: it works well enough to flag but not well enough to be definitive. Every institution I know treats the score as a starting point for conversation, not a conclusion. Your advisor has seen this before.