A small observation about Large Language Models (LLMs), human psychology, and some common pricing models for the APIs of hosted models: LLMs induce Gell-Mann Amnesia1 when asked about any topic about which the inquirer is unfamiliar. If one should ask an LLM a question to which one knows the answer, one scoffs at the small or large errors it makes. When one asks it about something with which one is unfamiliar, it outputs something "reasonable" that is difficult not to take at face value.
