Meet GPTZero: the AI-powered anti-plagiarism program created by Edward Tian, a computer science and journalism student at Princeton. GPTZero is a revolutionary program designed to combat academic plagiarism generated by the new AI-powered chatbot called ChatGPT, which has been stunning audiences with its ability to generate human-like text. GPTZero takes a unique approach to tackle the issue of academic plagiarism, allowing students and teachers alike to be confident that their work remains original and unaltered.
Table of Contents
Open Table of Contents
Introducing GPTZero
Edward Tian, a college student studying computer science and journalism at Princeton University, recently created an app called GPTZero to help detect whether the text was written by AI or a human. The motivation behind the app was to help combat increasing AI plagiarism. With the success of OpenAI’s chatbot ChatGPT, which has been stunning audiences with its ability to generate human-like text, the need for GPTZero has grown. From what the student shared on social media, it seems that the process that analyzes the texts to see if they were made by a machine or human has a high rate of I’m successful. GPTZero is doing a good job. In my initial experiment with this app, I had it read through a recent text conversation I had with a Chatbot. After a couple of seconds, the app told me that it was reading machine-generated text. I decided to have it read some of my recent blog text. After several seconds it accurately figured out that it was reading human-written text. If you feed the program some text with at least a few paragraphs, it seems to do a better job. To assess regardless of whether a text was written by a human or AI, GPTZero is scored on its difficulty level and burstiness — referring to how complex it is and how unpredictably it is written. To demonstrate GPTZero’s capacity, Edward Tian uploaded two videos contrasting the app’s evaluation of an article published in The New Yorker with that of a letter written by ChatGPT. It accurately concluded that they were respectively written by humans and AI. Due to its popularity, GPTZero crashed shortly after it was released due to unexpectedly high web traffic and currently displays a beta-signup page. While we are still far away from machines taking over all our thinking, GPTZero serves as a reminder that robots have become increasingly capable of writing like humans. Will humans maintain any cognitive abilities in the future or are robots going to do literally all of our thinking for us? Our future IQs look pretty grim.
How GPTZero Works
Edward Tian, a computer science and journalism student at Princeton, recently created an AI-powered program to combat academic plagiarism. Dubbed “GPTZero”, the program is designed to quickly and efficiently detect whether an essay has been written by ChatGPT — OpenAI’s new AI-powered chatbot — or a human. To analyze text, GPTZero uses metrics such as perplexity and burstiness. Perplexity measures how complex the text is, while burstiness measures how randomly it is written. This allows GPTZero to accurately detect whether an essay was written by a human or by ChatGPT. To demonstrate this, The app correctly identified that a New Yorker article and a ChatGPT letter were written by humans and AI, according to Edward Tian’s videos. In a tweet this week, the college student explained that his creation would “quickly and efficiently detect whether an essay is ChatGPT or not.”
I spent New Years building GPTZero — an app that can quickly and efficiently detect whether an essay is ChatGPT or human written
— Edward Tian (@edward_the6) January 3, 2023
As a result of high web traffic and no more available bandwidth, the site crashed, and now displays a sign-up for the beta version. In the course of my initial trials, I tried inputting ChatGPT’s ungrammatical text, as well as some of my own recent writing. Instantly it figured out that one was machine-generated. Then I tried my recent blog, and it was quick to tell the difference, even when I was holding the phone differently. The more you type into the program, the more accurate the results will be — so if you write out several paragraphs of content, you’ll have an accurate estimation. Will humans maintain any cognitive abilities in the future or are robots going to do literally all of our thinking for us? But for now, GPTZero is here to help us understand where AI-generated text ends and the human-generated text begins.
The Benefits of GPTZero
In the little time I’ve spent with it, GPTZero seems to work pretty well. When I fed it text from a recent ChatGPT chat, within seconds it determined that the copy was machine-generated. Similarly, when I fed it text from a recent blog of mine, I also realized that I was the human behind the prose. The better the accuracy, the more text you should plug into the program — for the best readout, at least several paragraphs should be analyzed the text to see how unpredictably it is written, as that’s how AI would write. His video compares the app’s analysis of a New Yorker article and a letter written by ChatGPT in terms of its complexity and burstiness. It correctly identified that they were respectively written by a human and an AI.
GPTZero has multiple benefits, including detecting instances of AI-generated plagiarism and helping educators identify students who may be cheating by using AI-powered chatbots. Furthermore, since its launch, the website hosting the app has crashed due to high traffic, showing that there is a high demand for such an application. Overall, GPTZero is a very useful tool for anyone looking to combat academic plagiarism generated by ChatGPT. Not only does it quickly and efficiently detect whether an essay is written by a human or machine, but it also has the potential to help educators keep students honest by recognizing AI-generated plagiarism.