The Science Behind
AI Detection
How do we know if a text was written by a human or a machine? It comes down to two key linguistic measurements: Perplexity and Burstiness.
1. Perplexity
Perplexity measures how “surprised” a language model is by the text.
AI writing chooses the most statistically probable words, resulting in low perplexity.
Humans make unpredictable word choices, resulting in high perplexity.
2. Burstiness
This measures the variation in sentence structure and length.
AI writing is monotonous. It uses sentences of similar length and rhythm.
Humans are chaotic. We mix short punchy sentences with long, complex clauses.
3. Watermarking
Advanced models like GPT leave invisible statistical footprints or “watermarks” in their token selection.
Our algorithm is trained on millions of samples to recognize these specific signatures that are invisible to the naked eye.
Visualizing the Difference
Imagine writing as a heartbeat. AI writing is a flatline—steady, perfect, and predictable.
Human writing is a spike—it jumps up and down. It has emotion, errors, and sudden changes in direction. Our tool scans your text and overlays these patterns to generate a probability score.
Try the Detector