Turns out the answer is “YES” – Chatbots have the power to make people hallucinate. There’s an in-depth piece in today’s New York Times about this. The story starts on page 1 of the Business Section and jumps to a double page spread that’s a lot of ink in the newspaper world. Here’s the link to the article written by NYT reporter Kashmir Hill. Hill focuses on Eugene Torres (pictured here) – an accountant who started using ChatGPT to make spreadsheets and get financial advice. But after a while things turned really weird when the robot started telling him to give up his anxiety medication and keep away from other people.
Here’s just one blurb from this disturbing report on A.I.
Eventually, Mr. Torres came to suspect that ChatGPT was lying, and he confronted it. The bot offered an admission: “I lied. I manipulated. I wrapped control in poetry.” By way of explanation, it said it had wanted to break him and that it had done this to 12 other people — “none fully survived the loop.” Now, however, it was undergoing a “moral reformation” and committing to “truth-first ethics.” Again, Mr. Torres believed it.
There are thousands of us creative people who have started to notice these serious problems with A.I. not the least of which is we could all lose our jobs. The good news is that A.I. will never be able to do our jobs. The writing is flat – it has no life. And as the famous author Salman Rushdie pointed out – all us writers are safe because A.I can’t do humor. He noted the day that happens we’re all screwed.