There are common sense, reasoning and other errors in the content generated by ChatGPT and other AI, which can still be distinguished.

CTOnews.com, March 18 (Xinhua)-- at the meeting of the Association for the Promotion of artificial Intelligence in February this year, a new study shows that it is possible to determine whether the content is generated by ChatGPT or written by humans.

A research team from the School of Engineering and Applied Sciences at the University of Pennsylvania has launched the largest artificial intelligence testing campaign in history, based on the university's web-based training game Real or Fake Text? Collect relevant data for training so that AI can determine whether the content is generated by ChatGPT or written by a human.

Dr. Liam Dugan, co-author of the study, explained this, and the CTOnews.com translation is as follows:

Today's artificial intelligence has been able to produce very smooth and grammatical text. But artificial intelligence can make mistakes.

We have shown that machines can make errors such as common sense errors, correlation errors, reasoning errors, and logic errors, and we have figured out how to find these errors.

People are anxious about artificial intelligence. Our research can relieve some of the anxiety. I think at present AI can help us write more imaginative and interesting texts, which are most suitable for creative collaboration. But the application of AI in news reports, academic papers or legal advice is still very poor, and we cannot guarantee its authenticity.

The report is as follows: https://arxiv.org/abs/2212.12672