Would also be helpful if we, and the author does not here, avoided use of inappropriately anthropomorphic concepts to describe LLM output, such as "hallucination". We already have a more appropriate word available, "confabulation".