A No Brainer

I’ve ignored my brain for too long.

A tiny twinge of pain in a hip and I’m off to see a physical therapist. My understanding of hips is pretty solid, and this understanding is built on formal PT-taught knowledge of elbows, knees, ankles, and to some degree toes, in my mental index.

Why? I have years and years of distance running in my history. My joints had to work, and sometimes they didn’t. I took my brain for granted when I ran, set it free to wander. I’m not sure why—it’s just always been there, chugging along, the little engine never failing to do for me the things I love to do.

The existential presence of mechanical brains has forced me to reckon with my own. Coming to grips with my own brain has been an emotional experience. Since I was a kid, I’ve never been able to reconcile a belief in life in another dimension like heaven or hell with my experiences growing up in rural central Illinois, now a bastion of MAGA. This is all we get so make the most of it.

A church my sisters took me to before I started school was spawned from a religion practiced in Appalachia from whence my father came. There was speaking in tongues, people falling to the ground, a loud ruckus which scared me. When I was older, my mother told me I was terribly rattled when they dropped me back in the holler. She laughed about it. “Who were they talking to?” I wanted to know.

So the question was framed. There is nothing religious about it, but what is this feeling inside me of a person, a whole self, a being? A self-delusion? How can these feelings wrapped up in an impossibly private history be just about a brain, a three pound hunk of meat we drag around with us, as philosopher John Searle describes it?

A huge help in this regard has been neuroscientist Suzi Travis, who publishes regularly on Substack and is a gifted writer and teacher. Suzi walked me (us) through arguments people schooled in this issue have had. Being a Substacker, I could comment, and Suzi responded.

I owe a debt to her for shaking me loose from my strong unconscious belief that human consciousness is somehow separate from the physical brain. I no longer think it is, and being on firmer ground makes me celebrate my brain as I do my heart. It was freeing to fight hard for my position but nonetheless give it up. I no longer resent my brain for being a brain and not the house of a spirit. Suzi’s work is incredible:

The Homunculus Fallacy and the Infinite Regress

***

If the technology industry had not invented language bots, I would have had no reason to open this can of digital worms. Somewhere during my AI-inspired journey to the center of the brain, which began during Thanksgiving, 2022, after I took the plunge and downloaded GPT3.5, my attitude toward my brain changed.

That first date with the bot was tortured—I was afraid that my bank account would be drained in the morning, that all of my passwords would be floating free in deep cyberspace. GPT3.5 failed miserably in its attempt to write a poem, my first ask of it, but I could see something was happening I’d never seen. My iPad never spoke back to me before. My feeling tone was omg what is this? Sort of like Close Encounters of the Third Kind.

I quickly got obsessed with figuring out what this machine was doing. It was crazy that everyone I read said “Nobody knows—even the machine doesn’t know.” I couldn’t find any humans to explain it to me, so I turned to the machine and asked it to teach me. Bottom line for me: The language bot is a language bot. It relies on letter strings, morphology, and syntax. A more apt moniker would be “artificial language.” In some ways, bot output is as devoid of meaning as the rote output of language from students who have memorized facts and vocabulary for a test.

My take on the phrase “artificial intelligence” weighed “artificial” heavily and equated the phrase with “non-real intelligence” and therefore not intelligent. I sometimes wondered how different the world might be if the Dartmouth Conference in 1957 had chosen a phrase like “mechanical language generator” to obscure any thought that bots would be intelligent.

In no way did I believe AI had human intelligence—still don’t believe. I was sublimely puzzled. Why was I so out of step? Within a week I’d drawn a firm conclusion: This bot is dumb as a fence post. My opinion of fence posts has risen since that Thanksgiving. Nonetheless, though I’ve seen language bots get smarter and smarter in the way people used to say to distinguish flip phones from smart phones, nothing fundamentally has changed. I don’t concede that bots can write, nor can they read, like humans, because literacy is a sociocultural phenomenon, not strictly a physical phenomenon as it is to bots. There is a sameness to bot output, a maddening redundancy, a total lack of taste and subtlety. Bot output is artificially human, about as nutritious as Diet Coke.

A big part of my changed attitude toward my brain stems from reading Suzi’s posts. I held some assumptions about my brain that I didn’t hold about, for example, my heart. I love my heart, always have, especially how it got me through 26.2 mile runs and, instead of weakening from abuse, grew stronger from use. I adore my mouth. It sings for me, it speaks for me, and it eats food. I’ll stop there. My brain? All I ever got from it was trouble. Things would be going along swimmingly and then boom! My brain would make trouble and I’d have to suffer through confusion.

My attitude was this. Don’t tell me this hunk of meat, as John Searle famously called it, hidden inside a skull, never seeing the light of day, rules my world. I am in charge of me, and I am not physical, no way. I am ethereal. In a way I resented this argument that all sense of life, including my spirit, is made up of atoms. Once I grasped the idea that brains are natural organs that produce consciousness, that bots are synthetic devices that produce language, I felt the disequilibrium subside. Just as hearts are pumps that circulate life-giving blood, brains are more than hunks of meat, brains circulate life itself through our bodies.

There is no such thing in my world as artificial life.

*****

Vygotsky’s book titled Thought and Language (1986, revised and edited by Alex Kozulin) underlay much of my initial unease over the word “intelligence” yoked with “artificial.” Clearly, thought for Vygotsky is separate from language. “Thought and word are not connected by a primary bond,” he wrote in Chapter 7, a lengthy argument pulling together his empirically grounded theory of inner speech developed throughout the text (p.211). “A connection originates, changes, and grows in the course of the evolution of thinking and speech.”

In this discussion Vygotsky is adamant, based on the fact that human infants think without words for the earliest months of their lives, that thought is the primary capacity linking humans consciously to the social and physical world. Learning about people and objects in the world is produced through cognition in interaction with the surround with no inner voice at all. Training, a word used loosely here, is self-motivated and self-regulated through wordless experiences. Training for language models is, as you know, completely linguistic. There is no thought involved at all, no search for a word to express a meaning. Bots learn through random trial and error. They don’t select words for the aptness of their meaning but for the probability that a word would appear in this linguistic space.

In humans the union of thought and language produces “verbal thought,” and the smallest unit of such thought is “word meaning,” which is not “an association between a word’s sound and its content” (p.213). Word meaning in fact has nothing to do with how a word sounds. As a unit of verbal thought, a meaning does not attach to particular individual words at first. Meaning pushes outward by way of thought and tries on words, using particular word meanings and senses, which are not stable and fixed but contextualized, only for a candidate words to push back on thought. In this pushing and pulling verbal thought leads to selection of a word.

In humans for Vygtosky, thought begins with meaning or sense, which can begin to activate a search for a particular word. Language bots begin and end with particular words and search for words that usually appear together. From there bots play the odds and dispense words that have been used before in similar linguistic situations. Any action even resembling the smallest unit of human verbal thought—the interdependence of meaning looking for words, meaning being reshaped and sharpened as candidate words emerge—never occurs in language bots.

Vygotsky rejects the notion that word meanings are fixed and stable. Implications for verbal thought are profound. “Let us consider the process of verbal thinking from the first dim stirring of a thought to its formulation” (p.217), a dynamic process that changes as the child develops. Meanings continue to evolve and change over time with individual interactions and experiences; therefore, verbal thought must always begin from scratch: “…[W]ords cannot be put on by thought like a ready-made garment” (p.219). The ready-made garment is made following the fluid, murky area where meanings and senses from thinking emerge in language.

The artificial nature of bots stems directly from their inability to engage in verbal thought. First, bots can’t think in the sense of spontaneously attending to a meaning apart from words and emotionally or intuitively sparking cognition. Quite capable of dealing in verbiage, prefabricated ready-made garments, bots are seductive, especially to learners who are ignorant of the interconnectedness of thought (human) and language (cultural)—and the understanding that bots do not engage in verbal thinking. “Inner speech is not the interior aspect of external speech—it is a function in itself” (p.249). Bots do not have inner speech.

Perceptive consciousness, the inner impressions and sensations that come together in thought to create a Gestalt, and intellectual consciousness, the consequence of verbal thought and external speech, of analysis and reflection, could be conceived as separate—two distinct types of consciousness. In humans, however, it is the blending of thought and language that unites human consciousness, the anchor in the physical world as a physical being. In Vygotsky’s terms:

“This aspect of the word brings us to the threshold of a wider and deeper subject, i.e., the problem of the relationship between words and consciousness. If perceptive consciousness and intellectual consciousness reflect reality differently, then we have two different forms of consciousness. Thought and speech turn out to be the key to the nature of human consciousness” (p.256).

Humans manage to bring together their sensory world and their intellectual world. This unified consciousness renders human existence unique, qualitatively different than other forms of conscious life. While language bots can participate in their mechanical way in intellectual consciousness, even with digital sensors, they cannot participate in perceptive consciousness. They have no skin in the game, and they have no thought. Humans bring two planes of consciousness together through verbal thought. Personally, I can’t see how AI will ever achieve this feat without becoming human.

…