When talking with LLMs, especially recent frontier models, what's frightening is that they speak more accurately than any expert.

It's embarrassing to admit, but I tend to trust the research materials LLMs bring more than my colleagues or the programmer friends I once respected.