Categories: AI Content

by Jonathan Gillham


How ChatZero Detect AI Generated Content

With the rise of artificial intelligence (AI) technology, it is becoming increasingly difficult to distinguish between human and machine-generated content. GPTZero, sometimes called ChatZero, is a powerful tool that seeks to address this problem. 

It uses advanced algorithms to detect AI-generated content in conversations, allowing users to identify and take action on any suspicious activity. This article will explain how GPTZero works and how it can be used to help protect online discussions from malicious bots and other automated accounts.


GPTZero is a language model designed to classify text as either human-written or generated by an AI model. It was trained on a diverse corpus of human-written and AI-generated text, focusing on English prose. The following are three features of GPTZero:

This model can give predictions on the sentence, paragraph, and document level

1. Multi-Level Classification

GPTZero allows users to determine how much of a document has been generated by AI technology. It does this through multi-level classification, which analyzes the text on sentence, paragraph, and document levels

On a sentence level, GPTZero can detect when AI or human writers have created portions of text. This allows for an accurate assessment of the amount of AI-generated content present in a given document. 

For documents that return mixed results, GPTZero provides a score indicating the probability that the entire document was written using AI technology.

Perplexity score

2. Perplexity Score

The Perplexity Score indicates how well a language model can predict a sequence of words. GPTZero computes this measurement for each document to determine the level of uncertainty the model has when trying to differentiate between human-written and AI-generated documents.

Measured on a scale, a low perplexity score tells us that the language model is confident in its prediction, while a higher perplexity score suggests that the model is uncertain. This metric provides valuable insight into how accurately the models can classify text as either human- or machine-generated.

Burstiness score

3. Burstiness Score

The Burstiness Score is a metric used to measure the repetition of similar phrases or topics within any given document. GPTZero provides this score for every text piece to indicate the similarity level between different sections.

The higher the score, the greater the chances that an AI model was employed to generate such text since artificially-generated content often includes multiple repetitions of phrases and ideas. Businesses, researchers, and organizations who want to identify whether machines or humans wrote certain texts can use this type of analysis.


  • GPTZero accurately detects whether a text is AI-generated or written by a human.
  • Can be used to identify potential risks of using AI in written work.
  • Works across various AI language models.


  • Accuracy on shorter documents is lower
  • Limited free credits

How to Use GPTZero

Step 1: Visit the GPTZero website

Visit the GPTZero website

Step 2: Enter the text you want to analyze in the text box provided

You can also upload a file

Step 3: Click the “Get Results” button

Don’t forget to read and agree to the terms of service

Step 4: GPTZero will analyze the text using its “perplexity” and “burstiness” indicators to determine whether a human or AI wrote it

Check out your document’s perplexity and burstiness scores

Step 5: GPTZero will highlight AI-Generated text as yellow

Free version allows you to get the result of the first 5000 characters

Comparing GPTZero to Other AI Platforms

To assess the accuracy of GPTZero’s content recognition system, we conducted a trial. We took sample content from Jasper.AI and ran it through GPTZero to see if it could identify that an AI (not a human) created the material.

# Title GPTZero
1 What is Jasper.AI? 80.05% 100%
2 How Does Jasper.AI Work? 41.48% 100%
3 Who Created Jasper.AI 44.70% 99%
4 Who is Jasper.AI for? 60.66% 92%
5 How much does Jasper.AI cost? 62.95% 62%
6 How to write an article with Jasper.AI? 69.46% 2.00%
7 Is Jasper.AI Free? 67.96% 99%


The table shows the results of GPTZero’s AI-content detection tool and’s tool to check the originality and uniqueness of various titles related to Jasper. The scores obtained from were consistently higher than those obtained from GPTZero, except for the sixth title, where GPTZero scored higher.

Although GPTZero shows perplexity and burstiness scores, it could not completely detect that a human did not write the content. Hence, GPTZero’s AI content identification tool is not as reliable for detecting the originality and uniqueness of content.

It also doesn’t provide a plagiarism-checking feature. This can be extremely concerning for companies that strongly rely on content produced by AI, as it could result in potential copyright infringement.

Advantages of GPTZero

Despite the underwhelming results earlier, GPT Zero has several advantages. The significant advantages it can provide are perplexity scores, which measure how well the model predicts the next word in a sequence, and burstiness scores measuring the likelihood of a group of words appearing in a document consecutively. 

Furthermore, GPTZero can detect a range of AI language models, including ChatGPT, GPT-3, GPT-2, LLaMA, and AI services based on those models.


All in all, GPTZero is an AI-powered tool designed to distinguish between human and machine-generated content. 

It offers multi-level classification to determine the amount of AI-generated content in a given document, perplexity scores to measure the model’s uncertainty in differentiating between human and AI-generated content, and burstiness scores to indicate the degree of similarity between different sections of text. 

While the accuracy on shorter documents is lower, GPTZero accurately detects whether a text is AI-generated or written by a human and works across various AI language models. The tool can identify potential risks of using AI in written work and help businesses, researchers, and organizations identify whether machines or humans wrote certain texts. 

However, it is unreliable for detecting content originality and uniqueness and does not provide a plagiarism-checking feature.

About the Author: 

Jonathan Gillham

Founder / CEO of Originality.AI

I have been involved in the SEO and Content Marketing world for over a decade. My career started with a portfolio of content sites, recently I sold 2 content marketing agencies and I am the Co-Founder of, the leading place to buy and sell content websites. Through these experiences I understand what web publishers need when it comes to verifying content is original. I am not For or Against AI content, I think it has a place in everyones content strategy. However, I believe you as the publisher should be the one making the decision on when to use AI content. Our Originality checking tool has been built with serious web publishers in mind!

Related Posts

  • AI writing tools are well on their way to becoming mainstream. ChatGPT only launched in November 2022, while GPT-4 was released in March 2023. As such, there is a great deal of interest in these AI tools’ ability to create fully-fledged writing pieces. But such innovation doesn’t come without ethical concerns. In the education space, […]

  • The future of artificial intelligence is here, and its name is GPT-4. Developed by OpenAI, GPT-4 is the newest and most advanced model in deep learning, capable of processing text and image inputs to produce human-like outputs.  While still not entirely on par with human abilities, GPT-4 has already achieved impressive feats in various benchmarks. […]

  • ZeroGPT is a tool designed to detect whether a given text is generated by an AI tool like ChatGPT or Google Bard, or if a human writes it. Users can enter their text and analyze it with the tool’s complex algorithms, which use DeepAnalyse™ technology to determine its origin.  The tool claims an accuracy rate […]