With the release of ChatGPT, many people including students have taken to the AI program to help or complete a task. However, now an app has been released to check what AI has written.
Created by Edward Tian in three days, GPTZero uses ChatGPT against itself to see what AI has produced.
The website is designed to recognize whether a body of text is written by AI or a human. It was announced in January, Tian aimed to create the artificial intelligence tool based on the analysis from the research of Princeton Ph.D. candidate Sreejan Kumar and the work of Princeton’s Natural Language Processing Group.
The free application is available to everyone, especially students to review student documents for AI fraud. The Terms of Service of the application mentions some fees for some services, with a 30-day free trial for new subscribers.
The accuracy of GPTZeros is disputed by many users on the internet, especially since the developer stated that the accuracy may not be 100%.
GPTZero can be used through its website here, where users can paste the text in the box, or submit a file that needs to be reviewed.
The AI program examines the use of text, known as complexity, and the consistency of this complexity within the text, known as burstiness. AI is very stable in its uncertainty and fallibility while humans can vary their writing.
ChatGPT has provided a free tool for people to misuse the service for training, and a tool like GPTZero was needed to ensure that the trust in the AI service does not go away.