Researchers from the University of Kansas Claim 99% Accuracy in Detecting Fake ChatGPT Conversations

Researchers at the University of Kansas have developed an algorithm that can identify academic writing done by AI bots with an accuracy of over 99%. As AI chatbot content continues to grow across industries, the ability to differentiate between robotic and human writing grows increasingly pertinent. Attempts to create ChatGPT detectors have not been successful so far, leaving the market wide open for new identification technologies. The new algorithm works in specific contexts and has identified particular characteristics of AI writing. The peer-reviewed paper that describes the method shows that the algorithm detected articles written by ChatGPT 100% of the time, while identifying 92% of AI paragraphs. The researchers hope that others will use their work to tailor detectors to their own niches and purposes. However, the algorithm was created specifically for scientific papers, and it remains to be seen how well it would perform in other contexts.