The new AI classifier for indicating AI-written text has been launched with the aim of distinguishing between text written by humans and text generated by artificial intelligence. This tool can be used to identify AI-generated content on various platforms.
The classifier has been trained to differentiate between human-written and AI-generated text from various AI providers. Its main purpose is to assist in preventing misinformation campaigns, identifying academic dishonesty, and distinguishing AI chatbots from real human users. Users can use this tool to assess the authenticity of the text they encounter.
While the classifier is not foolproof, it shows promising results. In evaluations on English texts, it correctly identifies 26% of AI-generated text as "likely AI-written," and mistakenly labels human-written text as AI-generated only 9% of the time. The tool's accuracy increases with the length of the input text and is more reliable with recent AI systems.
This classifier is available for public feedback and testing. The company welcomes input from users on the tool's effectiveness and is committed to continually improving the methods used to detect AI-generated text.
It's important to note that this classifier has some limitations. It should not be the sole method of determining the origin of text and may be less reliable for short texts. It's recommended for use with English text and may struggle with other languages. Very predictable text and edited AI-written text can also pose challenges for the classifier.
Overall, the new AI classifier offers promise for the detection of AI-generated content, but it's not foolproof. Users can provide feedback to help improve the tool's reliability. While it may not be completely reliable, it provides a starting point for identifying AI-generated content.