At the Association for the Advancement of Artificial Intelligence conference in February, the results of a new study showed that it is possible to determine whether content is generated by ChatGPT or written by humans.
A research team from the School of Engineering and Applied Science at the University of Pennsylvania recently launched the largest artificial intelligence detection activity in history, based on the web training game Real or Fake Text? created by the university. , collect relevant data for training, so that the AI can tell whether the content is generated by ChatGPT or written by humans.
Liam Dugan, Ph.D., co-author of the study, explains it as follows:
"Today's artificial intelligence can already generate very smooth, very grammatical text. But AI makes mistakes. We've shown that machines make mistakes like common sense mistakes, correlation mistakes, reasoning mistakes and logic mistakes, and we've figured out how to spot them. People are anxious about artificial intelligence. Our research may alleviate some of the anxiety. I think at the moment AI can help us write more imaginative and interesting texts, best suited for creative collaboration. But the use of AI in news reporting, academic papers or legal advice is still so bad that we can't guarantee its authenticity."
