Substantial spurious Information provided by ChatGPT

I’ve recently begun using ChatGPT, and am finding it both enjoyable and interesting.

Out of curiosity, I asked the AI for information about my late father-in-law, giving his name, profession and employer. He was a successful engineer and well-respected within his industry, but certainly was not more widely known. ChatGPT initially claimed that he had invented a device that was, in fact, the work of one of the founders of his company. A subsequent response accurately identified his title and true contribution, but also (falsely) claimed that he was the recipient of an MBE and other accolades, and gave inaccurate dates for his death and the age at which he passed away. When challenged, ChatGPT acknowledged that he had not received the awards it had previously claimed.

I fully understand that a) We are in still in the early stages of the AI revolution and b) my father-in-law was not a public figure and so I was stretching the capability of ChatGPT. However, whilst I’m very impressed that it found any information at all concerning my father-in-law and his work, I’m a little concerned that it also generated spurious information which it was unable to justify (due to mis-identification, or otherwise) when subsequently challenged.

If this observation is of any interest to the development team, I would be happy to provide more specific personal information to them via a private communication.