“An LLM is effectively a curated database, and methods of linking and updating the datasets accessible. The datasets are the result of ‘training’; structuring both the data derived and the ways it can be curated, linked, and accessed.

There IS an »aspect of one’s mind that slightly resembles a curated dataset, but only slightly. Because the ‘data’ isn’t mere information, tokens, or patterns. In fact, it cannot be this. Nonetheless, there’s a slight resemblance to an aspect of our minds… and that aspect can be imagined to be a structured, curated database.

But that aspect only functions, and poorly at that, in representational contexts. These are fake, though, for humans, consequential. It’s not trustworthy, and if you attempted to use »only that feature of your mind, you would die, quickly. First of misadventure (failure to effectively be aware in context). Additionally, it’s wildly untrustworthy.

Of course, in actual human minds, the ‘database’ and the ‘data’ are nonordinary. Extremely so. Probably infinitely so.

The resemblance is slight.

Do not confuse the two.

One has the potential for living intelligence and even transentience.

The other, literally, is a mechanically curated database.”

— infraheard