OpenAI Discontinues AI Detection Tool for ChatGPT-Generated Content
The arrival of ChatGPT sparked concerns about ethics, particularly in the education sector. People worried about students using ChatGPT to complete assignments undetected.
To address this issue, OpenAI, the company behind ChatGPT, developed the AI Classifier, a tool to identify AI-generated content.
But according to some other sources, OpenAI had stopped using the AI Classifier because of its poor accuracy.
OpenAI this decision in a blog post that they have decided to discontinue their AI Classifier tool as of July 20, 2023. The reason for discontinuation is cited as the tool’s low accuracy rate.
They acknowledged the need for improvement and stated they are working on better techniques to detect AI-generated text. OpenAI also intends to create tools for detecting audio and visual content produced by AI.
From the start, the AI Classifier faced doubts about its reliability. OpenAI admitted it was not entirely accurate. In evaluations, AI classifier accurately labelled 26% of AI-written content as “likely AI-written” but incorrectly 9% of human-written text as being produced by AI.
Though other tools are available, they are only somewhat reliable. As more people use generative AI like ChatGPT and Google Bard, the demand for better detection tools is expected to rise.
OpenAI might explore a revamped version of the AI Classifier in the future.