The developer of ChatGPT has given up on its tool designed to detect AI-written text, in a blow for efforts to combat cheating and disinformation.
OpenAI quietly dropped the program, called a classifier, because of its “low rate of accuracy”. Originally launched in January, the project was designed to combat “automated misinformation campaigns, academic dishonesty and positioning an AI chatbot as a human”.
The company acknowledged then “that identifying AI-written text has been an important point of discussion among educators”.
At the time OpenAI said the classifier “correctly identifies 26 per cent of AI-written text (true positives) as ‘likely AI-written’.”
The company did not give further details about how the classifier progressed, and posted an update to a blog, but given it has now been