We are definitely able to reason deliberately in ways that single/few-shot LLM inference calls cannot, which I think is somewhat obvious when you consider how m

We are definitely able to reason deliberately in ways that single/few-shot LLM inference calls cannot, which I think is somewhat obvious when you consider how much more complex our persistent brain is than a webgui for a chatbot, RAG or no. For more reading on what separates reason from understanding, I highly recommend (encyclopedia articles about) Kant & Hegel; they spent the better part of their lives musing on such topics. Long story short, reasoning is a process primarily oriented towards derivation instead of articulation. Of course, the next question is “are those higher level capabilities just made of many smaller stochastic parrots?” Unless you posit an ephemeral soul, I don’t really see how it could be anything else. What’s in the brain other than dumb neurons connected to each other? I suppose you could die on the stochastic vs. symbolic (deterministic vs. non-deterministic) hill, but that seems like pedantry at large scales, IMO.