OpenAI shuts down AI classifier
OpenAI cited accuracy concerns as the reason for discontinuing the tool.
In January 2023, OpenAI launched a free work-in-progress classifier that was trained to distinguish between AI-written text and human-written text. The company expressed its intention to obtain feedback on whether ‘imperfect tools like this one’ would be useful. With a focus on the impact of ChatGPT on education, OpenAI provided specifically developed ‘preliminary resources’ of information to educators. Due to the low rate of accuracy, the AI classifier has been ‘no longer available’ since 20 July 2023. OpenAI stated on its webpage that it will incorporate the feedback and assured its commitment to develop and use mechanisms to ‘enable users to understand if audio or visual content is AI-generated.’ However, OpenAI has not announced any plans to replace the AI Classifier tool with a more accurate version.
What was the purpose of OpenAI’s AI Classifier tool?
- The tool was a GPT model that predicted how likely it was that AI-generated a piece of text from various sources.
OpenAI developed the AI Classifier tool with the purpose of identifying AI-generated writing. Its primary objective was to detect instances of AI-generated content, which could be used for various purposes, such as misinformation campaigns generated by AI, engaging in academic dishonesty through AI tools, or falsely representing an AI chatbot as a human.