Most AI humanizers end with a claim: "Your text is now 100% human." ClearPen ends with two numbers measured the same way — the score before and the score after — and a sentence-by-sentence map of both. This post explains what those numbers are, how to read them, and why we think a measurement is worth more than a promise.
What Is the ClearPen AI Score?
The ClearPen AI score is the probability, from 0 to 100, that ClearPen's built-in AI detector assigns to a piece of text being machine-generated. It is reported at two levels:
- Document score — one number for the whole text.
- Sentence scores — a separate probability for every sentence.
Both are shown for your original draft and for the rewritten version, side by side.
How to Read the Colour Bands
| Document score | Band | What it means |
|---|---|---|
| 0–20% | Green — reads as human | The text reads as human-written. ClearPen will not charge to rewrite it unless you ask. |
| 21–50% | Amber — mixed | Parts of the text read as AI. The sentence heatmap shows which. |
| Above 50% | Red — reads as AI | Most of the text carries patterns associated with generated writing. |
Sentences are tinted on a gradient by their own probability, and any sentence at 50% or above is treated as leaning AI. That means a document can be green overall and still contain one or two sentences worth fixing — ClearPen tells you when that is the case.
Why the Before Score Matters as Much as the After Score
A single "after" number is easy to advertise and impossible to interpret. Was the text 90% AI before, or 10%? Did the rewrite help, or was the draft already fine?
ClearPen measures both, with the same detector, on the same text, in the same run. That gives you three things a lone score cannot:
- Proof of change. You see the difference the rewrite made on your own writing, not on a cherry-picked demo.
- A reason to stop. If your before score is already green, you do not need a rewrite — and ClearPen will not charge you for one. See fair-billing safeguards.
- A floor. If a rewrite would score worse than your original, ClearPen keeps your original. The after score never reads higher than the before score.
Why Sentence-Level Highlighting Changes How You Edit
A document score tells you whether there is a problem. The sentence heatmap tells you where.
In practice, AI-flagged text is rarely uniform. A draft is often mostly fine with a few telltale sentences: the generic opener, the tidy three-part list, the "In conclusion" wrap-up. The heatmap makes those visible immediately, in both the input and the output.
Click any sentence and you get its individual score with three options:
- Copy the sentence.
- Protect it, so it is kept word for word on the next run. See protected text.
- Spot Fix it — rewrite only that sentence, charged only if its score improves. See Spot Fix.
This turns humanizing from an all-or-nothing rerun into targeted editing.
Check your own text now
Paste a paragraph for a free AI-detection score — no signup.
What the Score Is — and Is Not
The ClearPen score is a consistent, transparent reference point. It is not a verdict about authorship and it is not a prediction of what any other tool will say.
That distinction matters because AI detectors disagree with one another constantly. They are trained on different data, use different thresholds and update without notice. Research has repeatedly shown that detectors produce false positives on human writing — especially formal, technical and non-native English writing. We cover the evidence in AI detection false positives and Is GPTZero accurate?.
What ClearPen can promise is that its own measurement is the same before and after, that you see it on your own text, and that it is shown for every sentence rather than hidden behind a single badge.
How the Score Fits Into a ClearPen Run
- Paste your draft. The before score and sentence heatmap appear within a few seconds.
- If the text already reads as human, ClearPen stops and asks whether you want to humanize anyway.
- Otherwise, the rewrite runs. The after score counts down from the before score when it finishes.
- Compare both heatmaps side by side and Spot Fix anything that still stands out.
See your before and after scores
Free plan, no credit card. Paste a draft and watch both scores on your own writing.
Frequently asked questions
- What does ClearPen's AI score mean?
- It is the probability, from 0 to 100, that ClearPen's built-in detector assigns to the text being AI-generated. ClearPen shows it for the whole document and for every sentence, before and after the rewrite.
- What score counts as human in ClearPen?
- ClearPen treats a document score of 20% or lower as reading human (green), 21–50% as mixed (amber) and above 50% as reading as AI (red). Individual sentences at 50% or above are flagged as leaning AI.
- Can I check my AI score without humanizing?
- Yes. ClearPen's free AI detector at clearpen.ai/ai-detector gives a document score for 10 to 5,000 words and needs no signup. Sentence-level highlighting is part of the ClearPen dashboard, including on the free plan.
- Will other AI detectors give the same score as ClearPen?
- Not necessarily. Detectors use different models and thresholds and often disagree on the same text. ClearPen's score is a consistent reference point measured the same way before and after, not a prediction of another tool's result.
- Why does ClearPen score the text before rewriting?
- So the improvement is measured rather than claimed, and so you are not charged to rewrite text that already reads as human. If the before score is already in the green band, ClearPen stops and uses none of your words.