I do think LLM seems to work similar to what the left hemisphere of the brain does. The left hemisphere deals with an abstracted world broken into discrete elements, and doesn't really make contact with the outside world--it deals with its system of representations. It also has a distinct tendency to generate bullshit, high suggestibility, and great respect for authority (which can apparently enter rules into its system of abstractions). The right hemisphere makes the contact with the outside world and does our reality checking, and it's really the more human element of us.

What this article says won't shock or disturb anyone deep into religious traditions with a strain of non-duality, which have had this message to shock and disturb people for thousands of years, in one way or another--there is no "you", especially not the voice in your head. I think you can come to a moment of intuitive recognition that the faculties of your brain that do reality checking aren't verbal, and they're riding shotgun to a bullshitter that never shuts up.

I think LLM can start looking more like automated general intelligence once it has some kind of link between its internal system of discrete abstractions and the external world (like visual recognition) and the ability to check and correct its abstract models by feedback from reality, and it needs an opponent process of reality-checking.