Guides

How AI detectors work, and what a humanizer changes

A detector looks for the patterns a chatbot leaves behind. A real rewrite takes those patterns out, and that is what moves the score.

What a detector measures

Most detectors do some mix of two things. They measure how predictable the text is: a language model picks likely words in a likely order, so very predictable text looks machine-made. And they run a classifier trained on piles of human and AI writing, which learns whatever separates the two.

In practice that comes down to the same things a careful reader notices: the stock words, sentences all built alike, paragraphs all the same length, and a tidy summary at the end. They are listed on the signs of AI writing page.

What a rewrite does about it

Slop Monkey rewrites the draft the way a person would say it. Each of these changes something a detector measures:

What the AI wrote

In today’s fast-paced world, remote work has fundamentally transformed how organizations operate. It is important to note that, according to Dr. Maria Chen, the share of employees working from home rose by 42% between 2019 and 2023, underscoring a pivotal shift in the modern workplace.

After Slop Monkey

Remote work changed how companies run, and it happened fast. Dr. Maria Chen found that the share of employees working from home rose by 42% between 2019 and 2023. That’s a big shift in a short time.

Why swapping words is not enough

The oldest trick is to run a draft through a synonym swapper. It leaves odd word choices in otherwise ordinary sentences, with the original structure still standing underneath. In 2025 Turnitin said its detector also looks for text that has been through that kind of tool. A full rewrite is a different thing: there is no original skeleton left to find.

Check the score before you copy

Next to every rewrite is a percentage: how human the result reads, from a detector we run on it for you. If you want it higher, press Go further and the rewrite is rewritten again. Every version is kept with its own score, so you can pick the one you like best.

Different detectors are trained on different examples, so their numbers differ from each other. If one in particular matters to you, run your final version through that one too.

Three things that push the score up further

  1. Go further. A second pass moves the text further from the original draft.
  2. Add something only you know. A number, a name, the day it happened. Specifics are what a model cannot produce and what a reader trusts.
  3. Make one edit of your own. Change a sentence or two after the rewrite. It is your text.
Paste your draft. 1,000 words free when you sign in.Start free