Burstiness
A measure of variation in sentence length, structure, and complexity across a piece of text.
Burstiness is a measure of how much variation exists in sentence length, structure, and complexity across a piece of text. Human writing tends to be bursty: a long, winding sentence followed by a punchy short one, then a question, then a list. AI-generated writing is often the opposite. Uniformly long, uniformly polished, uniformly mid.
Burstiness is one of the core signals AI detectors use. A piece of text with very low burstiness (high uniformity) raises the AI-probability score, all else equal. It's not definitive. Some careful human writers are also uniform. Which is why detectors combine burstiness with other signals.
Together with perplexity, burstiness forms the classic two-axis intuition behind many open-source AI detectors.
Related terms
- Perplexity· How surprised a language model is by a given piece of text. Lower means the text looks more model-generated.
- AI detection· The task of identifying text that was written by a large language model rather than a human.
- Stylometry· The statistical analysis of writing style. Historically used for author identification.
Move from definition to code
Free 1,000 requests/month, no credit card. Be detecting AI text in 5 minutes.