Artificial intelligence research startup OpenAI launched a tool on Tuesday to figure out whether text was written by a human or a computer.
ChatGPT creators, OpenAI, launch a tool to detect AI-generated texts
OpenAI, the company behind ChatGPT, a chatbot that can generate human-like text, has gone viral after it was released two months ago. Microsoft last week announced a multibillion-dollar investment in OpenAI and said it would incorporate the startup's AI models into its consumer and business products.
Since ChatGPT debuted in November, some of the largest school districts in the US, including New York City, have banned the AI chatbot over fears that students will use the text generator to cheat or plagiarize.
OpenAI CEO Sam Altman said education has changed in the past with the advent of calculators and other technologies, but he also suggested the company may have a way to help teachers discover written text from the AI.
OpenAI's new tool isn't perfect, and the company says it's still a work in progress and errors may occur while using it. “In our assessment of our 'challenge set' of English texts, our classifier correctly identified 26% of the text written by the AI (true positives) as 'probably written by an AI', while 9% of the human-written text was wrongly labeled as written by AI (false positives)," the OpenAI employee wrote.
The new tool is not reliable at parsing articles of less than 1000 characters, and OpenAI advises against its use for languages other than English. Also, the AI text may be updated slightly enough to prevent the classifier from correctly judging that it is not the work of a human.
OpenAI says it is reaching out to educators to discuss the capabilities and limitations of ChatGPT and will continue to work on detecting AI-generated text.
On offer on Amazon
