Peter,

Thanks for that link.  That cartoon is a precise characterization of how LLMs process data.  It was drawn in the 1990s when linear algebra usually meant something computed with matrices.  LLMs go one step farther by using tensors, but the results are in the same ballpark (or sewer).

Fundamental principle:  any machine learning system must be used with a system for evaluating or checking the answers.  For simple factual questions, a database can be used.  For more complex questions, logical deduction is necessary.  For any kind of system, ontology can detect obvious hallucinations, but ontology by itself is insufficient to detect incorrect details that happen to be in the correct category.

John

_____________________________________________