ChatGPT creators’ AI-written text detection tool: Here’s how to use it

OpenAI, the makers of ChatGPT, clearly heard the criticism of its AI chatbot. Namely, those of educators who have raised concerns about students using ChatGPT to do their homework or write their work for them.

On Tuesday, OpenAI released(Opens in a new window) a new AI classification tool that aims to address these concerns head-on. The free web tool was developed by OpenAI to help users determine whether a block of text was written by a human or a computer.

The tool is easy to use. Just visit the AI ​​classifier(Opens in a new window), log in and copy and paste the block of text you want to review. The classifier then ranks the text on a scale of either likely to be generated by the AI, possible, unclear, unlikely, or very unlikely.

Copy and paste text and find out if a human wrote it.
Photo credit: OpenAI / Screengrab

While this sounds incredibly helpful in addressing some concerns about AI-written text, OpenAI makes it very clear that there are caveats. For example, according to OpenAI, “the classifier is not completely reliable”.

“In our evaluations of a ‘challenge set’ of English text, our classifier correctly identifies 26% of AI-written text (true positives) as ‘probably AI-written’, while incorrectly identifying human-written text as 9% AI-written. of time (false positives),” the company said in a blog entry(Opens in a new window).

There are other limitations with the classifier addressed by OpenAI in its announcement. For example, the classifier works significantly worse for text that is less than 1,000 words. Sometimes it can mistakenly identify human-written content as computer-generated text. And it only works really well with English language text.

OpenAI isn’t the first tool to release a tool that helps detect when text is computer-generated. In fact, it’s not the first classification tool the company has released itself. However, according to OpenAI, it is “significantly more reliable” than its previous tool. And it is very remarkable that such a tool is also continuously being worked on by one of the leading minds in the field of AI.

However, there is still a long way to go before teachers and educators can confidently rely on an AI detection tool to uncover classroom fraud.