Exactly… People are conflating the ability to parrot an answer based on machine-levels of recall vs actually understanding something and being able to articulate how the machine itself arrived at a conclusion (which, in programming circles, would be similar to a form of “introspection”). LLM is not there yet