Menu

Stay Connected

Follow NFT Magazine

© 2022 NFT Magazine
All Rights Reserved

Latest News

ChatGPT Creators Release New Tool To Detect AI Writing

Picture of Joey Sendz

Joey Sendz

ChatGPT Creators Release New AI tool to detect AI writing

A new tool to identify text authored by AI has been released, according to developer of ChatGPT OpenAI. According to a statement, the business has unveiled a new classifier that will supposedly enable users to “discern between language produced by a person and text authored by AIs.”

The rapid growth of ChatGPT has demonstrated both the remarkable capabilities and the limitations of the technology. With this implementation, OpenAI is demonstrating that it is looking for solutions to increase the functionality of its ChatGPT system.

The creator of the popular chatbot phenomenon that has a multitude of sectors terrified for their jobs is OpenAI. The system’s powers have astounded people. It has demonstrated the advancements made by technology under the direction of the development team.

Related: Spotify Improving User Experience For NFT Holders

Elon Musk is one of the main founders behind OpenAI and ChatGPT
Elon Musk is one of the main founders behind OpenAI and ChatGPT

Now that the ChatGPT creators have published a brand-new tool to detect AI-written content, they are pushing the technology even farther. The developed tool will function as a “excellent classifier,” according to a statement from OpenAI, and “may inform mitigation for misleading claims that AI-generated material was written by a human.”\

The program is now “not fully reliable,” according to the paper, which also warns that more testing needs to be done. On the other hand, according to their most recent task set, it “correctly identifies 26% of AI-written text (true positives)”. Additionally, “incorrectly labeling the human-written text as AI-written 9% of the time.”

“Our classifier’s reliability typically improves as the input text length increases,” continued OpenAI. The most recent version of their classifier, they note, “is significantly more reliable on text from more recent AI systems,” when compared to their previously released classifier.

Related: AI And Blockchain: Making or Taking Jobs?

According to the research firm, the classifier is being made available to the public in order to promote program feedback. Furthermore, they added that they would like to know if these tools are helpful and that they would keep working “on the detection of AT-generated text,” in their endeavors.