why do models like ChatGPT forget things during conversations or make things up that are not true?
In: 803
Because ChatGPT is NOT A TRUTH MODEL. This has been explained from day 1. ChatGPT is not “intelligent” or “knowledgeable” in the sense of understanding human knowledge. It is “intelligent” because it knows how to take natural language input and put together words that look like a response to that input. ChatGPT is a language model – it has NO ELEMENT IN IT that searches for “truth” or “fact” or “knowledge” – it simply regurgitates output patterns that it interpret from input word patterns.
Latest Answers