Humans that produce output like LLMs are most likely to be diagnosed as schizophrenic, which I don't believe is the goal.
The problem is we are so used to yielding to confidence we don't apply the necessary checks even when we know it is projected by machine.