ChatGPT Creator Releases Anti-Cheating Tool for Teachers

Image: OpenAI

ChatGPT creator OpenAI today announced the launch of a new “classifier” designed to detect machine-written text.

“We’ve trained a classifier to distinguish between text written by a human and text written by AIs from a variety of providers,” the company said.

While the company’s AI-powered chatbot has taken the world by storm, it has also raised concerns about machine-generated text being exploited for schoolwork, misinformation campaigns, and more. Last week, ChatGPT made it into the news for passing the final exam for the Wharton School’s Master of Business Administration (MBA) program.

OpenAI said that its new text classifier tool is designed to address instances where AI-generated text may be passed off as written by a human, including for automated misinformation campaigns, academic dishonesty, and misrepresenting an AI chatbot as human.

To check whether a chunk of text was written by an AI, users can simply copy it into a box and the system will evaluate it. It will rate the text on a five-point scale, ranging from very unlikely to have been AI-generated to unlikely, unclear, possible, or likely. You can click here to try the classifier for yourself.

According to OpenAI, the tool works best with text samples that are greater than 1,000 words. how similar the text being analyzed is to the types of writing OpenAI’s tool was trained on.

OpenAI described the classifier as a “work-in-progress,” noting that it is not fully reliable. “In our evaluations on a “challenge set” of English texts, our classifier correctly identifies 26% of AI-written text (true positives) as “likely AI-written,” while incorrectly labeling human-written text as AI-written 9% of the time (false positives).”

P.S. - Like our news? Support the site with a coffee/beer. Or shop with our Amazon link. We use affiliate links when possible--thank you for supporting independent media.