Skip to content
Redmoon Calculators
Text analysis

AI Text Pattern Detector

Free AI writing pattern checker. Scores text on sentence length variance, burstiness, transition-phrase repetition, hedging, structural uniformity, and n-gram repetition. Does not claim to detect AI — surfaces measurable patterns commonly associated with machine-generated text.

Built and maintained by Paul Clark, Redmoon Software

When to use this

Use it as one signal when reviewing flagged text — not a verdict. AI detectors that claim definitive accuracy are unreliable; this tool gives you measurable patterns you can investigate, with full transparency about what each metric means.

How it compares

Most AI detectors are black-box services that report a single confidence number. This tool shows you each underlying metric so you can decide what matters for your context.

0 chars

How it works

The detector measures six linguistic signals — burstiness, uniformity, transition-phrase rate, hedge rate, 3-gram repetition, and vocabulary repetition — then weights them by the writing-mode preset.

Burstiness is the standard deviation of sentence length divided by the mean. Human writing is bursty; AI writing tends to flatten.

It does not claim authorship — only that the text contains patterns commonly associated with machine-generated writing. Treat it as one signal, not a verdict.

FAQs

Does this detect AI-generated text?

No. It surfaces measurable linguistic patterns commonly associated with machine-generated text. Reliable AI authorship detection is not currently possible.

What patterns does it measure?

Sentence-length burstiness, transition phrase rate, n-gram repetition, hedging language, and vocabulary repetition. AI writing tends to have low burstiness and high transition-phrase use.

Why is burstiness important?

Human writing alternates short and long sentences ("bursts"). AI writing tends to produce sentences of similar length.

Can it be fooled?

Easily — paraphrasing or human editing changes burstiness. Use the score as one signal among many.

Worked example

Input

In conclusion, it is important to note that machine learning has revolutionized many industries. Furthermore, it can be argued that artificial intelligence will continue to play a pivotal role in our digital age.

Output

AI-pattern score: 72 / 100 — Many AI-style patterns.

Both sentences use formulaic transition openings ("In conclusion", "Furthermore"), heavy hedging ("can be argued"), and similar sentence lengths. The transition-phrase rate alone pushes the score into the high band.

Common pitfalls

  • Does NOT detect AI authorship — only patterns associated with it.
  • Skilled editors easily rewrite text to lower the score.
  • Short samples (< 50 words) produce unstable scores.
  • Some academic and technical writing legitimately scores high on uniformity.

What this actually measures — and what it cannot

This tool looks for surface patterns associated with machine-generated prose: stock transitional phrasing, formulaic hedging, a particular register of evaluative adjective, and structural tics like tidy triads and summary sentences that restate the paragraph just given.

It is important to be clear about the limit. This is not a detector and cannot be one. There is no reliable way to determine whether text was machine-written from the text alone — published detectors have high false-positive rates, and they are documented to misfire disproportionately on writing by non-native English speakers, whose more formulaic constructions look like the patterns being flagged.

Treat the output as a style observation about the text in front of you, never as evidence about its author. Using a score like this to accuse someone is not a use this tool supports, because the underlying inference does not hold.

The legitimate use

What it is genuinely good for is editing your own drafts. The patterns it flags are worth removing regardless of who wrote them: "it is important to note that", "in today's fast-paced world", paragraphs that open by announcing what they will say and close by repeating it. That prose is flabby whether a person or a model produced it.

Used that way the score is a proxy for something real — how much of the text is scaffolding rather than content — and cutting the flagged constructions almost always tightens the writing.

Read more about this

Related tools

Send feedback

We read every message. Tell us what could be better or what you love.