The API helped me a lot with coding and logical problems. Recently I started to ask questions about a few of my favorite books. However, the answer is not related to the books asked: the API might create fake stories or summaries of a few books. For instance, I asked about the content of a story in the Grimm’s Fairy Tales: the louse and the flea. The answer got from chatGPT is not close to the content of the story. I tried to ask chatGPT three times about this, and I got 3 different versions of a story about a louse and a flea. I wonder if it is possible to add more training about storybooks. BTW I faced a similar issue when asking about the content of another book: “the silent cry” by Kenzaburo Oe. Some books might have no problem, such as “L’Étranger” by Albert Camus.
It was the same for me, except that in my case it was about the poem in a modern neighborhood by the Portuguese poet cesario verde, everything I asked him about that poem the chat gave me false answers like when I asked how many stanzas and verses the poem had and the same He answered me 53 verses and 4 stanzas and it has 20 stanzas and 5 verses each and in total it has 100 verses.