What AI can and can not do is highly dependent on the expertise and patience of the user.
For example, the made-up/fake citations are eliminated by a simple step that asks the AI to check the validity of all of the citations.
Malwarebytes tried to get ChatPT to write ransomware. If politely declined, but with persistence, replies, and persistence, the Malwarebytes author got it to do things it wasn't supposed to do. The story of how it shifted from a position that stated it could not and would not do this task ended up with the AI producing much of what the 'hacker' asked it to do. https://www.malwarebytes.com/blog/news/2023/03/chatgpt-happy-to-write-ransomware-just-really-bad-at-it
Similarly I've seen some very poor AI pictures, but I've also seen YouTube examples of AI videos to accompany music. Having been around for the release of songs like White Rabbit and Stairway to Heaven, I have to say, the AI videos were appropriate for psychedelic rock. Jerry
***********************************************
The BEE-L mailing list is powered by L-Soft's renowned
LISTSERV(R) list management software. For more information, go to:
http://www.lsoft.com/LISTSERV-powered.html