AI, Gods and Selves: Incredibly Effective Illusions

The overall thesis is good but the details are iffy. The biggest issue is that the bicameral mind can't have broken down midlife in adults during a single generation of humans in the late Bronze Age collapse; genes and brains do not work that way. (Similarly, Tor Parsons claims that schizophrenia is an exhibition of bicamerality, which also doesn't line up with genetics of schizophrenia.)

I think that a reasonable way to save the thesis starts with the mind-projection fallacy, from a different Jaynes. When a human anthropomorphizes, they project a perceived mind onto the anthropomorphized object. A person holding a religious token can have particular neural pathways which are activated by the perception of the token and also project a mind onto the token; their brain is where the token's corresponding deity is simulated, but perceptually the deity emanates from the token. This can neatly explain how people feel when interacting with chatbots; a projected mind in their brain takes on the responsibility of attributing emotions and inner thoughts to a pile of linear algebra, but the mind was always in the brain and didn't manifest physically in the computer. Following Hofstadter's logic, the self is a symbolic token whose activation causes the brain to anthromorphize the brain itself along with the rest of its attached body; humans are only anthropomorphized and given minds with a perceptual glance rather than as a matter of logical necessity, a priori knowledge, or inevitable thermodynamic substrate.

This Hofstadter self-token is precisely what could have been invented in the late Bronze Age! It could have spread memetically, slowly, through processes that are now incommensurate with today's science, prior to societal collapse. It could have even been (part of) the inciting event which caused so many Mediterranean societies to collapse, although archeological evidence suggests that there were pressures on food and freshwater supply, which would have been more dire. One big issue with this theory is that people in the earlier Bronze Age had first-person pronouns, discussed themselves, and showed off a rudimentary theory of mind; the self-token already existed. Another big issue, which many bicameralists omit, is that this is all extremely specific to the Mediterranean and doesn't generalize to contemporaneous Vedic or Chinese societies, so this probably wasn't about the invention of the original alphabet/abjad.

Ultimately I think that this entire video would have done better with a sharper thesis statement that picks a more psychoanalytic posture; the archeological evidence just isn't strong enough for this thesis. If they had said that chatbots are Echo and humans are Narcissus then the video would have been much shorter and easier to parse. I suppose that they would have also gotten many more low-quality replies arguing that, while most humans are Narcissus, the replier is actually a modern-day Apollo who has mastered the nymph chatbot and cannot be bested in any sort of divine tournament vibecoding competition.

Paragraph five! We could imagine crafting tokens specifically to activate certain neural pathways. The cell phone is an obvious attempt, designed for addictiveness, but we could probably be more humane about the whole affair. I'm reminded of an old Dresden Codak comic where the protagonist attempts to craft a token which will show her visions of the future:

What if the whole of my memory is stored solely in the material world, my mind nothing more than a reflection of events locked in the most incidental of things? How many more memories remain hidden?