> Alright, so we have our first point of divergence. For me, the majority of my general knowledge comes from reading

Reading is part of your environment - you didn't read your way into a vacuum - the very tools you use to read right now (language, symbols, syntax) are environmental gifts. Something, even if you're consciously unaware of it, made you read those material. The specific script you read is part of your environment (it's unlikely you will know what symbols in Hindi mean if you didn't grow up in that environment or weren't biased towards it). Your physical brain structure depends on the languages you speak and will physically morph as you change those languages.

If you have absolutely no clue what "mutton paratha" is, you won't ever start reading about "mutton paratha" right away. (there's another thread we can discuss about human biology, original thought and serendipity but that is better done over email than in HN comments). If you can hear and see, a lot of your general knowledge comes from hearing and seeing too, even if you're consciously unaware of it or don't acknowledge it. If you saw your dad working on cars all your young life, you might not think you have any idea about cars but you might be surprised.

> Where does the majority of an individual's _understanding_ come from?

Understanding is more complex than knowledge. Understanding and knowledge are at different levels of abstraction but you cannot understand something you have no knowledge of. Thus knowledge is a pre-requisite to understanding but insufficient. With sufficient understanding, you can synthesize knowledge. Nikola Tesla synthesized knowledge about AC Induction motors - a knowledge that's now taught even in high schools from his understanding of EMF. However given the same knowledge of EMF, many people at the times of Tesla didn't have sufficient understanding or motivation or capacity to birth AC Induction motors.

You can be extremely knowledgeable but understand absolutely nothing. LLM's are, I argue, excellent examples of this - they have tremendous knowledge but depends on human intervention and discourse to tease it out (as great as SOTA LLMs are, they are not going to randomly design apps without an initial prompt which provides them agency and biases them towards autonomy).

(Again, I'm discarding some details that we can discuss over email. For example: I can strongly argue that LLMs have some understanding of what its knowledge is - like the semantic distance between "mutton" and "paratha" and how they can change across various manifolds in latent space)

If we agree on this take on understanding, then I argue that the majority of an individual's understanding comes from their consciousness, when they have exercised those neural networks and processed their knowledge to arrive at inferences and re-enforced (belief) or disassociation (misbelief) of assertions

We agree on that take on understanding, yes. More succinctly, I think we could say that understanding comes from an application of, and interaction with, knowledge. The more connections to a bit of information that an individual has, the better their understanding.

This thread's getting a bit long, so I'm happy to continue this discussion via email, if you like. My email is ben@overmyer.net.