Meet GPTZero: An App Created by Student that Detects Essays Written by AI

It was November 2022 when Microsoft-backed OpenAI launched its superhero called ChatGPT and within a few months, it was unanimously concluded that the future of Generative AI had arrived and how. ChatGPT burst into the scene like a wondrous tool that was waving its magic wand and making lives simpler across all domains and industries.

But, as is the case with most technological innovations, this creation, despite its unbelievable admiration, too was met with a lot of apprehensions, criticism, uncertainties, and plagiarism concerns. Many had their reservations, especially teachers and leaders from the academic industry.

Recommended: Google’s Bard Now Has Programming Capabilities, Can Assist Users to Code, Debug & Explain 

ChatGPT – Fears of Plagiarism in Academia

From writing poems and college essays to generating marketing content and creating an app, this little chatbot was creating text on a variety of subjects and topics depending on the user’s prompts. So basically, this chatbot was becoming every student’s tech-savvy best friend.

A professor from Northern Michigan University, while assessing essay assignments came across an immaculate piece with perfect arguments and examples. Impressed with the finesse, when he confronted the student about whether the essay was written by him or not, he admitted to having used chatGPT. This was not one odd case. More such incidents were reported across different universities.

An article published in New York Times stated that many school and colleges, who were aware of how some students were using ChatGPT to write assignments, restructured their curricula. Take-home essays were replaced with in-class assignments or oral exams.

GPTZero – the App that Detects AI-Written Essays

A 22-year-old student from Princeton University has come up with a logical solution – a tool called GPTZero that detects any text written by ChatGPT. Created by Edward Tian created the app during his winter break and launched it in January.

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

Within the first week, the tool, crafted for academicians and educators was used by over 30,000 people, Emma Bowman of NPR stated. GPTZERO has partnered with Unesco, Canvas, and K16 Solutions to improve the tool.

What Does It Do?

Plagiarism Check: You can get a holistic picture of how much AI has contributed in a write-up.

Highlights text written by AI: After uploading a file, the app detects the sentences written by AI and highlights them.

Multiple file uploads: You can upload multiple files at one go, meaning, you can upload your entire classroom’s file all at once.

Easy API Support: The app has a simple-to-use API. Some of the featured API users include Salem, ISB, Port Angeles School District, Galvanize, etc.

Integration Support: Engineers are available to help users to customize the endpoints according to their requirements.

Recommended: 10 Simple Ways To Earn Money Using ChatGpt

How GPTZero works?

To determine if the text was written by a human or AI, the app takes into account two variables – the complexity, and the variation. Typically, texts which are complex in nature with different lengths of sentences are written by a human and the ones that are more uniform are written by AI.

Here’s a demo:

here’s a quick demo with john mcphee’s “frame of reference” pic.twitter.com/WphxfxxFdr

— Edward Tian (@edward_the6) January 3, 2023

The tool is undoubtedly very useful for professors and educators, but at the same time, blind trust can lead to believing false flags.

Edward believes that ChatGPT’s ability to be utilized for plagiarism is unlikely to be affected until the app is adopted by a significant company. He further stated,

“It’s always an arms race between tech to identify synthetic text and the apps. And it’s quite easy to ask ChatGPT to rewrite in a more personable style … like rephrasing as an 11-year-old. This will make it harder, but it won’t stop it.”

Toby Walsh, Scientia professor of artificial intelligence at the University of New South Wales, isn’t too convinced. He strongly believes that unless a major company decides to adopt GPTZero, it is unlikely to have an effect on ChatGPT’s ability to be used for plagiarism.

Final Thoughts

Tian explains that he has nothing against artificial intelligence and in fact, supports its potential, however, he wants to establish more transparency when anyone uses it. The future of GPTZero looks promising, but a lot of improvements with enhanced capacities to detect user text need to be made to put up a good show against ChatGPT.