QUESTION: You said that the German Nazi Party was raising money selling bonds in the United States before they invaded Poland in 1939. When I asked AI, if the Nazis sold bonds in the US it said “No, the Nazi regime did not sell sovereign bonds in the United States after coming to power in 1933 and before the outbreak of WWII in 1939.” So, who is correct? You or AI?
ANSWER: From what I am being told, a problem is surfacing with ChatGPT-generated content, which often contains factual inaccuracies. The development of language models to engage in AI is presenting a problem. They are learning from the WEB, correct. However, they are not necessarily capable of verifying what is true or false. Here is a Conversion Office for German Foreign Debts $100 Bond (Nazi Government sold in the United States) into the New York 1936. I have the physical evidence that suggests that the answer you received was incorrect.
British Journal of Educational Technology (BJET) recently explained that “no research has yet examined how epistemic beliefs and metacognitive accuracy affect students’ actual use of ChatGPT-generated content, which often contains factual inaccuracies. ” For those unfamiliar with this arcane term of philosophy, linguistics, and rhetoric, epistemic, it traces back to the knowledge of the Greeks. That Greek word is from the verb epistanai, meaning “to know or understand.”
I try to be accurate, and if I state something as fact, I have generally verified it versus making a statement of just an “opinion,” perhaps derived from a belief. Nobody is perfect – not even ChatGPT.
