How AI Detection Works
This tool uses statistical and linguistic heuristics to estimate the probability that a piece of text was generated by an AI system like ChatGPT, Claude, or Gemini.
The 6 Detection Signals
- Sentence Uniformity: AI tends to write sentences of very similar lengths. Humans vary more.
- AI Transition Phrases: Phrases like "it is worth noting", "in conclusion", "delve into" are disproportionately common in AI output.
- Vocabulary Diversity (TTR): The ratio of unique words to total words. AI reuses a smaller set of common vocabulary.
- Passive Voice: AI models default to passive constructions more than human writers typically do.
- Word Sophistication: AI tends to prefer longer, more formal vocabulary on average.
- Punctuation Diversity: Human writers use dashes, semicolons, and ellipses more freely; AI output is often more uniform.
Accuracy Note: No heuristic AI detector is 100% accurate. Highly formulaic human writing (academic papers, legal documents) may score high, while well-edited AI output may score low. Use this as an indicator, not proof.