Comment by claritise

Just create a RAG with wikipedia as the corpus and a low parameter model to run it and you can basically have an instantly queryable corpus of human knowledge runnable on an old raspberry pi.