I use AI to compile lists of current bestsellers in different Hispanic countries for me. Also, I ask it to find books similar to certain other books. About 1 in 8 suggestions are completely invented. Non-existent books, imaginary authors.
Last week I asked it to find a scholarly article on a certain subject. It gave me a list of five and the first two were messed up. I didn’t look at the rest because I felt I was wasting time.
Another problem is that it ignores my prompts. For example, I say, “I need a list of bestsellers in Peru, Spanish-language only, not translations.” The result I get starts with two translated books, and the AI acknowledges it. “I know you said you didn’t want translations but here are two excellent novels that are translated.”
It’s fine, it’s still helpful enough. I did find an interesting Peruvian novel with AI help yesterday. But the point I’m making is I thought AI would start substituting doctors because it can diagnose. But this amount of invented crap among helpful things? You can’t have that in medicine. It’s not occasional, it’s systematic.
I don’t understand how it works at all, so maybe it’s that I’m simply not getting it.
Clarissa's Blog 
Leave a comment