With the advent of AI, a number of applications and services have emerged that are having a profound impact on our lives. Now that advancements in AI technology are spreading rapidly, everyone is looking for shortcuts and answers to fairly simple questions on the Internet to “save time”. People in academia, in particular, are using online services like AI chatbots to find quick solutions, address plagiarism, and more.
Now, with the advent of such platforms Chat GPT by OpenAI, which has been trained to generate long-form responses, many people are using it to spin well-referenced and unusual naturalistic subjects. Although OpenAI said it would watermark ChatGPT output, people like Edward Tian are determined to solve this problem of AI plagiarism for good.
I spent New Years building GPT Zero — an app that can quickly and efficiently detect whether an article is ChatGPT or human-written.
— Edward Tian (@edward_the6) January 3, 2023
Over New Year’s break, a computer science student at Princeton developed an app called GPT Zero. As the name suggests, this app can “quickly and efficiently” determine if your subject is a ChatGPT creation. Tian showed off his creation in a series of tweets where he showed examples of GPT Zero in action. One of the examples shows that the New Yorker article, “Frame of Reference,” was written by John McPhee while the LinkedIn post was created by a bot.
GPT uses “precocity” and “burstiness” to determine the text written by the zerobot. Confusion, in machine learning, refers to how surprised a language model is when it sees new data, while the tendency for a term to be repeated once in a document is called bursting. Is. Human writing has more fluency which leads to more variation in sentences, and language models can better predict text when the fluency is low.
As Tian points out in a Substake Newslettermore than 10,000 people have tested the publicly available. Version of GPTZero on Streamlit As of today they also mentioned that they have updated the GPTZero model to reduce false positives. Additionally, he is working on more updates to GPTZero including new capabilities and scalability.
Although Microsoft has plans to use ChatGPT’s capabilities to its advantage, the New York City (NYC) Department of Education has blocked access to ChatGPT on school devices and networks due to concerns about piracy.
Source: Polygons
