My students are often surprised to learn that LLMs aren’t answering their questions. Rather, an LLM answers the question “what would a reply to this look like?” It’s one of the first things I explain in the “Should I use LLMs?” portion of my syllabus.
@mcnees In my opinion a meaningless distinction. Answering a question, for human or bot, is the act of determining what a reply to this would look like (best case scenario). Both human and bot are capable of generating something that sounds like a reasonable reply but is wrong. To see this in action, ask a toddler to explain "why" and you'll get confabulations about "because."
@escarpment @mcnees why did you choose those words, rather than other words that had a different meaning but would also look like a valid reply?

FWIW what the LLM actually said in response was "That’s a sharp and important framing [...blah blah blah]".
@adam @mcnees I chose those words because that's what the determinstic system of my brain and fingers landed on at that point in time based on the prompt.