WROITER’s Blog
A curated blog about detecting & improving AI writing.
This isn’t a blog in the scroll-until-you-find-something sense. These guides are written in sequence and build on each other. Start at the top and work down — or jump to the question that brought you here.
Jump to your question
"I'm reading something and it feels off. How can I tell if it's AI-generated?"
The visual tells, the false leads, and how to verify a suspicion without overreacting.
How to Spot AI Writing →
"I need a process for checking AI writing. What's the right workflow?"
Five steps from sample collection to decision, with every step explained.
How to Check AI Writing →
"How do AI detectors actually work? Why do they disagree with each other?"
Perplexity, classifiers, pattern-matching, and why different bets produce different numbers.
How AI Detectors Work →
"Can I trust the result of an AI detector enough to act on it?"
When detectors are reliable, when they fail, and the right way to treat output as triage.
Do AI Detectors Work? →
"Has anyone actually been wrongly flagged as AI while being a human writer?"
Yes. Canonical texts and student essays can be misread. We show what is benchmarked today and what is directional evidence.
False Positive Hall of Fame →
"I just want to check if my text is AI generated."
Paste it, read the flags, check the context, decide.
Run the Diagnostic →
Or start from the top
01
02
03
04
05
How to spot AI writing
The pattern guide for reading with your eyes. Strong tells, weak tells, and the false leads that trip people up. Start here if you've ever read something and thought "this feels off" but couldn't explain why.
How to check AI writing
The full workflow: collect enough text, inspect the patterns, run a transparent diagnostic, compare against context, then decide. Five steps, in order, designed to keep you from jumping to conclusions.
How AI detectors work
The three main approaches — statistical, classifier-based, and pattern-matching — explained without jargon. Why two tools can disagree on the same text, and what that disagreement actually means.
Do AI detectors work?
The short answer is "sometimes." The longer answer explains when they're reliable, when they fail, and why a score should never be treated as proof of anything. The reliability brief you need before building a review process.
False Positive Hall of Fame
A case file on detector false positives: where human writing gets flagged, what we can publicly prove today, and how to read scores without overreaching.
Need to go deeper?
The blog answers the questions people search for. The method explains how the diagnostic works. The research publishes the evidence behind it.
Interpretation rule
A higher score means stronger overlap with AI-typical surface patterns. It is not proof of authorship, cheating, or intent, and it is never enough to skip human review. For score ranges and the full interpretation boundary, see How It Works. If policy is involved, read Limitations first.