Imagine that you ask a question of 5 high quality diverse LLMs, & they all give the same answer, & also seem confident in their answers. On average, what is the chance that their common answer is actually wrong?