Blogging websites have recently been invaded by ChatGPT-written articles. They are everywhere. It’s worse than the plague!
I see such articles everywhere — and sometimes I can tell just by scrolling through them that they were ChatGPT generated. However, sometimes, that may not be so easily noticeable. To solve that, I will show you the best tool you can use to detect ChatGPT written text and teach you how to get the best results. This will help you keep ChatGPT bot-writers out of your publication, or just get to know if a writer is actually good at writing, or just using AI. Let’s get into it!
ChatGPT Detector
Let’s get this detection tutorial started by generating some text using ChatGPT. Let’s see, how about “Tell me what Deep Learning is”. That should be a good start. Here is the result:
Also in text form:
“ Deep Learning is a subset of machine learning that involves the use of artificial neural networks with multiple layers to model and solve complex problems. These neural networks are designed to recognize patterns in large and complex datasets, and use these patterns to make predictions or decisions.
In traditional machine learning, the features or inputs that are fed to the algorithm are typically handcrafted by a human expert. In contrast, deep learning algorithms learn features directly from the data, which makes them more versatile and able to handle complex and diverse inputs.
Deep learning has proven to be highly effective in a wide range of applications, including image and speech recognition, natural language processing, autonomous vehicles, and more. The use of deep learning has led to significant advances in fields such as computer vision and artificial intelligence, and has the potential to revolutionize many other areas of science and technology in the future.”
This should be big enough to be tested. Now, you can go to the ChatGPT output detector website by OpenAI to test the text above.
I know it says it’s a “Demo” — however, the results are incredibly precise. Let’s put the prompt above and see if it detects the text as fake.
Well, would you look at that — the tool predicted with a 99.98% accuracy that the text is ChatGPT generated (aka. “Fake”).
Important Tips
There are some things you must consider when using this tool, to make sure the result is accurate.
Firstly, you need to use enough text, so you can make sure the algorithm has enough data to detect the correct Output. It usually works well even with less text most of the time, but you will see that the accuracy of prediction is reduced.
As you can observe above, the prediction accuracy was reduced by reducing the text. But it still predicts with high accuracy that it’s fake text.
However, if you take that text and edit it, and restructure it by hand, you may actually be able to fool the detector!
But you need to do it by hand because using other AI paraphrasing tools will not actually change the result. In the image below, you will see the initial ChatGPT text paraphrased by Quillbot AI paraphraser.
As you can see, even after Quillbot paraphrased the text, the result is the same. That means that you cannot get away with copy-pasting AI-generated content. Using OpenAI’s ChatGPT Output Detector anyone will have the ability to detect ChatGPT written articles and text!
Conclusion
Now you know how to detect if a text was written by ChatGPT. Hope this will help you in detecting ChatGPT text, no matter what the use case may be.
In my opinion, ChatGPT should be a tool to help writers be more efficient, not a way in which the Internet and blogs will be crowded with AI-written text and AI-written books.Any writer should use it — but not use it to write 100% of the article. Chat GPT should help them when they have writer’s block or when they can’t put what they think in words.
Hopefully, you are of the same opinion!
