AI detection glossary
Plain-language definitions for the terms you'll meet in AI detection. For developers, educators, editorial teams, and procurement. 16 terms, more added as the field moves.
Detection
AI detection
The task of identifying text that was written by a large language model rather than a human.
Confidence score
A 0-to-1 value reflecting how certain the detector is that a piece of text was AI-generated.
Sentence-level scoring
Returning an AI-probability score for each individual sentence in a document, not just one number for the whole thing.
Watermarking
Embedding a statistical signal in AI-generated text that detectors can later check for.
Statistics
Burstiness
A measure of variation in sentence length, structure, and complexity across a piece of text.
False negative
When a detector misses AI-generated text and classifies it as human.
False positive
When a detector flags human-written text as AI-generated.
Perplexity
How surprised a language model is by a given piece of text. Lower means the text looks more model-generated.
Stylometry
The statistical analysis of writing style. Historically used for author identification.
Models
Hallucination
When an LLM generates output that is fluent but factually wrong or fabricated.
Humanizer
A tool that rewrites AI-generated text to evade detection. And the arms race that comes with it.
LLM (large language model)
A neural network trained on huge text corpora to predict the next token. The engines behind ChatGPT, Claude, Gemini, etc.
Token
The smallest unit of text a language model processes. Usually a word or a piece of a word.
Workflow
Try AI detection in code
Definitions are useful. A working API is better. Free 1,000 requests/month.
Get my free API key