Ramblings on Consciousness, AI, and the Bicameral Mind

O, what a world of unseen visions and heard silences, this insubstantial country of the mind!

Over winter break, I read an extraordinary book by psychologist Julian Jaynes titled "Origins of Consciousness in the Breakdown of the Bicameral Mind." Jaynes makes the fantastic assertion that human consciousness is not merely a biological phenomenon of firing neurons, but that it came into being around the 19th century BCE, paralleling the development of language, metaphors, and cultural evolution. This would mean that the ancient Greeks were not conscious! That consciousness is a cultural phenomenon instead of a biological one! Arguing that consciousness is merely the very tip of the iceberg of habit, assumptions, beliefs, and instincts, he writes "it is perfectly possible that there could have existed a race of men who spoke, judged, reasoned, solved problems, indeed did most of the things that we do, but were not conscious at all."

Jaynes believed that humans exist in two separate spheres - the physical, object-oriented world, and the mental world of decisions, fears, introspection, and consciousness. He argues, quite effectively, that in the psychology of early civilized humans, these two spheres were kept separate, and only through the development of language and decentralized cultural power did we develop the ability to inhabit both at the same time. Crucially, he cites language as one of the most influential factors in creating consciousness. THAT idea is the crux of my fascination with Jaynes - the thought that words, linguistics, and natural language hold the key to unlocking a deeper understanding of human consciousness. Cormac McCarthy's essay on the Kekule problem further emphasizes the role of language on creating the conscious mind, citing combinatorial grammar as the mechanism that enables the uniquely human gifts of poetry, prose, mathematics, and temporal transcendence.

Fascination with linguistics has been informing philosophers, psychologists, and neuroscientists for years. It is now time for artificial intelligence researchers to join the fold. Arguably the biggest problem in the AI field is that of value alignment - instilling human values and emotions into machines to ensure that their goals and directives do not harm humans. For example, computers that learn the human values of kindness, selflessness, and honesty would present much less of an existential threat to humans as such computers become more powerful and essential to human existence. Natural language processing (NLP), a form of machine learning, can "read" human texts and draw conclusions from them. In a recent study by the University of Vermont's Computational StoryLab, machine learning algorithms were able to detect the emotional arcs of novels. Clearly, language holds as much power for machine consciousness as for human consciousness. Just as our ancient ancestors may not have been conscious in ways that we can understand, artificial intelligence may soon become conscious in ways that could threaten our race's existence. Not only does language hold the keys to understanding human consciousness, but it could also allow us to shape machine consciousness in a way that benefits humanity.