How GPTZero scores your text
GPTZero reads a document and gives it an overall classification (human, AI or mixed) along with a probability. It also highlights individual sentences it thinks are most likely machine-written. Teachers usually see both: the headline number and the highlighted lines.
Under the score are two ideas GPTZero has talked about publicly since it launched: perplexity and burstiness. Newer versions combine these with other signals, but they are still the clearest way to understand why a paragraph gets flagged. If you fix both, the rest usually follows.
Perplexity and burstiness, explained simply
Perplexity measures how surprising your word choices are. AI writers like ChatGPT, Gemini and Copilot pick the most likely next word again and again, so their text is very predictable. A sentence like "Technology plays a crucial role in modern education" has low perplexity; every word is the one you would guess. People make less obvious choices, use specific nouns and occasionally phrase things a little oddly.
Burstiness measures how much that predictability varies from sentence to sentence. Human writing is uneven. A long sentence with two clauses and an aside is followed by a short one. Then maybe a question. AI output tends to keep every sentence at a similar length and rhythm, which looks flat on a chart and suspicious to a detector.
Other things that push the GPTZero score toward AI:
- Stock phrases such as "delve into", "it is worth noting", "plays a pivotal role" and "in today's fast-paced world".
- Transitions at the start of every paragraph: "Moreover", "Furthermore", "Additionally", "In conclusion".
- Paragraphs that follow the same template: claim, explanation, neat summary.
- Generic examples with no names, dates, places or numbers.
How to get a human score on GPTZero
You can fix this by hand or with a humanizer. Most people do a bit of both.
- Clean the text first. Copying from a chat window can bring invisible characters and markdown symbols. The free AI text cleaner removes them in your browser.
- Paste the text into The Humanizer above. Pick Enhanced mode if the original scores strongly as AI, or Standard if it is already mixed.
- Choose the purpose that matches your writing (Essay, Email, Blog/Article or General) so the tone fits.
- Click Humanize. The rewrite keeps your meaning, facts and citations, and varies sentence length and phrasing so the perplexity and burstiness look human.
- Click Check AI score to compare the original and the result side by side.
- Read it once and add a detail only you would include: a personal example, a source you actually read, a specific number.
If you prefer to edit manually, focus on the highlighted sentences. Break up long, even runs of sentences. Replace generic words with specific ones. Cut transitions you do not need. Manual editing works, but it is slow on anything over a few hundred words, which is where the tool saves you time. For a broader guide, see how to make AI text undetectable.
Check your score before you submit
Never submit or publish a draft you have not checked. It takes a minute and avoids an awkward conversation later.
- Run the final version, not an earlier draft, through the AI detector. Small last-minute edits can change the score.
- Look at the sentences that still score high and rewrite just those.
- Check the whole document, not only one section. GPTZero scores the full text, and one heavily AI paragraph can drag the overall result.
- Read it aloud. If it sounds like you, it is usually fine. If it sounds like a press release, keep editing.
Different detectors weigh signals differently. If your school also uses Turnitin, our guide on how to bypass Turnitin covers what it looks for.
Why your own writing can get flagged
GPTZero is not perfect. Careful, formal writing can score as AI simply because it is very regular. This happens a lot to non-native English writers, who often stick to safe grammar and common vocabulary, and to anyone following a strict essay template. If that is you, humanizing your own text with Standard mode adds natural variation without changing what you said.
A note for students: you are responsible for following your school's academic-integrity policy. The Humanizer is best used to polish your own drafts, or AI-assisted drafts, into writing that reflects your own voice and thinking.
Free users can humanize 200 words per run, 5 runs a day, with no sign-up. For full essays, the Basic and Pro plans allow 1,000 or 3,000 words per run.