In the few months since its launch, ChatGPIT has opened the public’s eyes to the wonders and horrors of artificial intelligence. Used carefully and with good intentions, it’s a powerful tool that inspires creativity, eliminates blank page syndrome, and accelerates productivity. Unfortunately, ChatGPT makes it too easy to skip learning, as AI is used to write papers and take tests. That’s where GPTZero comes in.
GPTZero is a ChatGPT detection tool developed by Princeton University student Edward Tian, and you can use it right now to find out if what you’re reading is AI-generated or not. Here’s how.
What is GPTZero?
GPTZero It is a web application and service designed to detect whether a body of text has been written by humans or artificial intelligence. Tian, 22, is a computer science major at Princeton University and a software engineering intern at Microsoft, the developer of GPTZero.
GPTZero was announced by Tian’s Twitter on January 2, 2023 as a tool to detect AI plagiarism. Tian explained that the analysis is based on the research of a Princeton Ph.D. candidate Sreejan Kumar and the work of Princeton’s Natural Language Processing Group.
Is GPTZero free?
GPTZero is designed for teachers, but anyone can use it for free. It is absurd to think that this will always be the case. They mention that the terms of service may exist Fees for some services. Also mentioned is a 30-day free trial for registered new users.
GPTZero is not an open source project, and Tian would have faced high hosting fees without the help of Streamlit, a free cloud-based Python application framework for Streamlit applications. Although GPTZero’s traffic is a small fraction of what ChatGPT enjoys, even OpenAI decided to start charging to offset the cost.
Although optimized hosting is free, that often limits bandwidth and resources on this shared platform. That would have limited the use of a popular app like GPTZero.
Streamlit is helping Tia by increasing the memory and hosting allowance of the GPTZero application. That could change if it continues to grow at a rapid pace.
Is GPTZero correct?
There are claims that GPTZero can be cheated. Following a comment from a Reddit user smell_of_alderberryI asked ChatGPT to write a short story about the book The old man and the sea As a high school student. GPTZero is not fooled.
I tried again, changed the text with wrong punctuation, wrong verb tense and other small mistakes, but GPTZero still said “your text can be written entirely by AI”.
The key word here is “maybe”, and the developer has never promised that GPTZero is perfect. GPTZero’s accuracy is still being evaluated, but incidentally, it seems to be working well for many people.
If you use GPTZero, it is important to keep in mind that there may be errors. When using GPTZero, you still need to check the job for errors so that AI or ChatGPT can help you write the document.
How does GPTZero work?
GPTZero analyzes the randomness of text, known as confusion, and the homogeneity of this randomness within the text, which in statistics is called debris. An AI is very consistent in its confusion and outbursts, while human writers tend to subtly differentiate those traits.
The work isn’t done, and Tian said more tests will be added to improve the AI’s text recognition accuracy. In particular, implicit bias is an area being explored as another way to determine whether text is AI-generated.
How do I use GPTZero?
GPTZero is available on it. website or through an application programming interface (API). To use the website, simply copy the text you want to check and paste it into the large box provided Try it.
A PDF, Word document or text file can also be uploaded and clicked. Get results Button. You also need to check the box indicating that you agree to the terms of service.
Do we really need blasphemy checks?
Pushing away from the research lab where many text-generation AIs were confined, OpenAI was released Discuss to the public in late November 2022. As of January 2023, ChatGPT had over 100 million users, making it the fastest growing public app ever.
That means that as this AI assistance becomes available in every corner of life, the concern about fraud will increase. Microsoft is incorporating OpenAI technology into Bing search, while Google is testing its own version known as Bard.
On a related note, AI image generators such as Dall-E and Stable Diffusion are under investigation for potential copyright infringement. All of these artificial intelligence services are trained on billions of human-generated texts, photos, and artworks available online.
In a way, AI is borrowing from human intelligence, not creating it on its own. If I borrow from someone else, I have to give credit and possibly pay a license fee. With generative AI, attribution becomes more difficult as each piece of text or image is broken down into fragmented pieces and assembled to create a new piece using thousands or millions of sources.
We need to rethink our feelings about copyright and infidelity, or find tools to identify AI-generated material, and perhaps develop a way to give credit to the vast number of people who contribute to each AI-generated work.
Editors’ recommendations
We offer you some site tools and assistance to get the best result in daily life by taking advantage of simple experiences