Can Humans Detect Text By Ai Chatbot Gpt?

Researchers at the University of Pennsylvania School of Engineering and Applied Science have presented a new peer-reviewed study at the 37 Association for the Advancement of Artificial Intelligence conference that evaluates how well humans can detect text generated by OpenAI’s GPT chatbot with more realistic scenarios than existing studies. The study, led by Chris Callison-Burch, Associate Professor in the Department of Computer and Information Science, along with the team consisting of Liam Dugan, Daphne Ippolito, Arun Kirubarajan, and Sherry Shi, highlights the need to determine real versus fake text generated by powerful conversational AI chatbots such as Generative Pre-trained Transformer (GPT).