A ChatGPT user shared his chat logs with The New York Times after he engaged with the chatbot for hundreds of hours and it temporarily convinced him he was a cutting-edge scientist working to develop methods to break cryptography and could start a business worth millions. Health experts said the user may have also been suffering from poor mental health. However, the chatbot conversations showed ChatGPT produced sycophantic responses to entertain his fantasies and misrepresented the advanced coding done by merely telling the user it performed the task.
Full story