I see this take a lot but I am not sure its entirely true. If we were to blindly ask questions based on a models world knowledge, yes there will be false statem

I see this take a lot but I am not sure its entirely true. If we were to blindly ask questions based on a models world knowledge, yes there will be false statements. When asking questions or summarizing data across a provided dataset, I have not seen this to be a problem. Now I think there could be an interesting question on how its summarizing, there may be inherent biases but I don't believe the original issue is in fact a problem.