AI Writing Heuristic Checker
Paste text and see the countable patterns that tend to show up in machine writing, each one shown with its raw value.
Wondering if this text is AI written? This checker counts the surface patterns that tend to appear in machine prose and shows you every number behind the score, so nothing hides in a black box. It measures em-dash density, which makes it a quick em dash AI checker, alongside stock vocabulary, sentence-length variation, rule-of-three lists and hedging. You paste your writing and a 0–100 likeness score updates as you type. This is a transparent, heuristic approach, not an AI detector verdict: a high number means the text carries countable tells, never that a person did not write it.
Read the guide: How to Spot AI-Written Text (and Why You Can't Be Sure)Your text
Nothing is uploaded. Every count is worked out in your browser as you type.
AI-likeness score
88
Many AI-writing tells · 54 words, 3 sentences
16 flagged words per 1,000 (delve, robust, leverage, seamless…).
2 em-dashes (—) per 1,000 words. Models lean on them heavily.
Variation in sentence length (lower = flatter, more machine-like). Humans vary more.
2 three-item lists per 1,000 words (a, b, and c).
2 hedges per 1,000 words (typically, generally, it's worth noting…).
Words per sentence. Long, even sentences read more machine-like.
This is a heuristic, not a verdict
A high score means the text carries countable patterns common in machine writing. It is not proof anyone used AI. These signals disappear the moment text is edited or "humanised", and plenty of careful human writing trips them too.
No AI detector is reliable. The best-performing tools top out around 85% accuracy and still flag genuine human writing as machine-made, which is why schools and publishers warn against acting on a score alone. Treat this as "here is what is countable in the text", never as an authorship judgment.
How it works
- 1
Paste your text
Drop an article, an email or a model answer into the box. Every signal recomputes on each keystroke, entirely in your browser, so you can edit and watch the numbers move.
- 2
Read the likeness score
The big number blends six weighted signals into a 0–100 estimate of how machine-like the writing reads. Higher means more of the countable tells are present.
- 3
Check each signal
Under the score, every signal shows its raw value and a one-line meaning: em-dashes per 1,000 words, stock vocabulary hits, sentence-length variance, rule-of-three lists, hedging and average sentence length.
Instant & 100% private — nothing is uploaded
Every calculation runs locally in your browser. The prompts, token counts and numbers you enter stay on your own device and are never sent to a server — nothing is stored, logged or shared.
Frequently asked questions
- Can this prove text was written by AI?
- No. It counts patterns that are more common in machine writing, but every one of them also appears in ordinary human writing, and they vanish the moment text is edited or reworded. Treat the score as a description of what is countable, not a judgment about who or what wrote it.
- What does the score actually measure?
- It blends six signals: em-dash density, stock AI vocabulary, sentence-length variance, rule-of-three lists, hedging phrases and average sentence length. Each has a fixed weight, and each is shown with its raw value so you can see exactly why the number came out where it did.
- Why do human writers get flagged too?
- Careful, formal writing uses many of the same devices: balanced lists, longer even sentences, words like comprehensive or crucial. That overlap is why no honest checker can separate human from machine on style alone, and why this tool never claims to.
- How accurate is this compared to a real AI detector?
- Even the best commercial detectors top out around 85% and still flag genuine human writing as AI, which is why schools and publishers warn against acting on a score alone. This tool is deliberately simpler and more honest: it shows the raw counts and makes no accuracy claim. Do not use it to accuse anyone.
Important
For planning and estimates only. Prices come from a published rate table dated on the page; providers change pricing without notice, and token counts here are approximations. Confirm against the provider’s own pricing before you budget or commit.
More tools
More from the Hivly network
Free sister tools on our other sites.
