Finding Our Place in the Age of Robot Gods

Maintaining our humanity during the inevitable arrival of robot gods. Remembering that bread alone is insufficient.

What is our purpose in an age of rapidly advancing intelligence?

I struggle with this myself. I’m a product manager, close to AI/ML tools, but also someone battered by the information + notification overload. During leisure times I'm overwhelmed with options on YT’s faux TikTok "short" section and at work I'm overwhelmed by Slack and Teams messages.

I am equal parts thrilled about the potential of having the aggregate of human knowledge at our fingertips as I am gravely concerned that the current age of narrow AI glory and the Artificial General Intelligence hype cycle will wreak havoc over human society. Our economy, our institutions, our livelihoods, and sense of purpose will all rattle as the Promethean ambrosia enters its final stage of god like ability. The problem is not the technology itself, but how we will choose to deploy it.

I certainly share the existential risk concerns of Nick Bostrom and others, who want to guard against us becoming servile puppets of an intelligence greater than our own, and have specific thoughts on how to pursue the computation behind governance and oversight. However, of equal and perhaps more immediate concern is our ideological approach to current technology, institutions, society, etc. The shape of organizations reflects the products and impact they have on our minds and lives. The problem is that we generally have a poor model of human behavior, i.e., one that is overly reductive.

The Stench of Reductionism

We value output as the end in itself. We do not value the human for what we can broadly describe as their own moral, psychological, spiritual depths. We do not adequately chart the challenges and journey of a person; we only measure the goods they buy along the way. We proxy social value with money, we proxy psychological states with, at best, happiness measures (but more often, again with money) and measure social value in GDP terms (sensing a common theme?).

I view this as at least a partial consequence of reductionism, a notion I borrow from Dennis Noble in his treatise against Dawkins' Selfish Gene perspective but that I feel applies equally well in various other domains of thought from economics, psychology, linguistics, business and our institutions more generally.

Philosophers have long chased golden calves in objectively defining good and bad, beauty and ugliness. But from Wittgenstein’s language games to Zen riddles, the lesson is the same, their boundaries fall into the changing river.

Academically, the tendency to model complex processes using single variables or proposing fixed relationships among phenomena is what drives the interest in things like the selfish gene among Neo-Darwinists and the selfish agents that maximize utility using prices among neoclassical economists. It's the tendency to reduce human behavior to biological sex and thereby minimize human agency or conversely, to maximize human agency in promoting telelogy (Adler) or even to define a fixed set of archetypes applicable across human cultures (Jung).

AI currently has a conscious feel. Despite the statistically correlative nature that forms the backbone of the most noteworthy models, many are able to pass some version of Alan Turing's litmus test for consciousness. Meaning they can successfully fool us into thinking they are an actual person. And if the Turing test is too thin a measure, that should frighten us more, not less.

These machines CAN get both iteratively and categorically smarter and more capable. A machine that matches how living systems, like say, a primate child, can extrapolate and ultimately form abstract representations by training on a relatively sparse amount of data that has access to a large language model's API, i.e., an in-silico agent with our cognitive capacity to form sparse models but unconstrained by our cognitive limits, will convert us from the cowboy to the horse.

So, we're screwed? Hopefully the matrix is pleasant?

The reductionist chef offers his patrons the same chicken seasoned with a single ingredient. The approach to both AI and AI risk both have the stench of reductionism lingering like a poorly cleaned cast iron pan. As neoclassical economists reduced current and future economic value to prices and humans to self-seeking agents, Neo-Darwinists reduced the complexity of life to genetic code and the selfish gene. Deep learning is likewise reducing the complexities of our language using a cost function. We do not simply use language to arrive at an answer we have backed into our heads but use language to discover the right question and what we feel while seeking an answer out loud, on the page, or in our heads.

This same dynamic is at play in AI Risk. Given the immutable advance of AI capacity driven by LLMs, other traditional underlying AI tech and more advanced approaches that may build on a model of cognition more advanced than the verifiably more reductive and data-hungry approaches of current LLMs, the future of our place in the universe is a little more questionable. The future of work is, in one sense, as uncertain as it's ever been.

EXCEPT if you follow the zeitgeist and acknowledge our responsibility for creating just such an intelligent entity. I.e., we seem to be allocating the bulk of our national 401K, our aggregate GDP and the insights of our best scientific minds towards the next and most robust AI models.

The problems

How can we align a machine to human values and have it work towards our aims without a good understanding, or model, of human aims?

What does it mean to be human at a time when robots can synthesize our best minds? When our work, our best effort at carrying ourselves and society forward, lies in sacrificing any hope of productive output to the seemingly inevitable out-shadowing of our intellectual capacity by LLMs?

While many are correct in emphasizing the need for AI governance, many organizations that cover this topic fail to model the underlying human. 80,000 Hours, Scott Alexander, and many other thought leaders in the realm of AI existential risk tend to approach humanity in a reductive fashion, not unlike how LLMs reduce human intelligence to physical patterns of output. They reduce humans to, say, a reported measure of happiness and define the cause as monetary value.

Why is art as a psychological tool relevant?

It's critical that we understand the forces at play when we respond to a survey saying we're happy, or answer why our life feels unfulfilled, etc. This is critical as this technology will be shaping our future. My approach to an answer, i.e., how I would model a solution, lies largely in our psychology. Specifically, I believe it lies in Jungian-style individuation and his emphasis on individuation. Most poignantly, in Viktor Frankl's use of logotherapy. The most ancient tools we have for elevating our psychological states are the same ones we continually employ today: art.

Art provides meaning for individuals and groups by enabling a certain form of structure through an organic cultural construction process. When J.R. Tolkien painstakingly translated Beowulf he was placed directly in the seat of the protagonists psychological struggles, so moved to create his own modern narrative. Art thus enables yes, group cohesion, but cohesion that involves the necessary aggregation of moments in an individual's life across time and within a broader social, and what Jung would call, "cosmic" context. The most poignant example of this is in one of the more ancient art forms, at least one that has left us with tangible historical artifacts: the epic.

Narrative and drama remind us that we operate in a world beyond our control yet where our decisions have consequences. In other words, they enable us, through an impulsive ability to empathize with a given protagonist, to intuit the structure that we are part of, that we need to access virtues to enable our better selves and overcome challenges and conflict.

Our Cosmic Moment

The tidal wave of machine intelligence that seems to be barreling towards us is certainly a unique moment in history, but in an important way there's historical precedent that can inform us at this critical socio-cultural juncture: the rise of literacy.

Joseph Henrich marks this as a culmination of various religious and cultural forces that make western, or "WEIRD," societies unique. The rise of Lutheranism, in Henrich's telling, promoted literacy in order to disintermediate priests and the greater papacy in religious life through individualizing understanding of the Bible and other Christian teachings.

This phenomenon brought fuel to the rocket engine of western civilization. It developed an important trajectory that led to the moment we're now in, one that began with Ancient Greece and the forces of scientific understanding and ran throughout the Roman effort to construct a society based on the Rule of Law.

This had many profound consequences, including an impact on art in a way that's informative for our moment in time. The most important form of art for societies until then was embodied in what we think of as epic poetry. It brings together a list of individual psychologically potent art forms in a way that serves both individual and group cohesion. The epics of Gilgamesh, Beowulf, and ancient Anglo-Saxon poems tell a common story among a particular culture using rhythmic, metrical, and narrative patterns that resonate with listeners and give meaning to mythologized or historical events.

Medieval peasants knew their place in the cosmic order through embodied rituals and oral traditions. Literate Protestants had to rebuild that cosmology using ink and individual interpretation. We're facing a similar reconstruction, but instead of disintermediating priests, we're potentially disintermediating human cognition itself.

What impact did the rise of literacy have on this?

Some good and some bad. Firstly, why have a psychologically attuned set of content to transcribe cultural narratives if everything can be written out in a book or downloaded online? The trouble is we have access to our history yet find it boring. The road that led to a disparate society reduces art from its religious and spiritual tethering, when the best art was the stuff of mythology, when the Bible and Hesiod's Theogony were revered and not just enjoyed as pastime. Postmodernism and the troubles of modern life likely begin in our ability and tendency to devolve away from the common and resonant tools that once bound us individually and collectively. This can be seen in poetry and music in particular.

While Western music created a strong tradition and perhaps culminated in what we call classical music as exemplified by Mozart, from the time when Pythagoras tried to structure music by adopting mathematical ratios between sound frequencies, the use of music has been overly reduced by a reductive-scientific oriented mind. Not a dynamic and psychological tool capable of adapting to our complex lives and the often ineffable experiences we have of them.

However, from another viewpoint, Western society offers cultural tools that close the gaps. We have substituted for the yawning lacuna left by our epics. Firstly, there is the rise of the novel. Though we live in a world dominated more by prose than poetry, a likely reverse of the cultural environment preliterate societies inhabited, novels offer a way of giving poetry to prose. For individuals now trained in reading an "objective" recounting of events in a dry, dispassionate, factual manner, they can do so now in a way that lets the epic story, the passionate challenge of the protagonist, or the clever manner of the trickster reemerge into a seemingly staid existence.

Think of Virginia Woolf's Orlando. The narrative is presented in a dry manner, a faux biography that maintains a frame that gives it the feel not of a Chernow biography but one even more dedicated to a factual sequence of events. Yet suddenly we learn the protagonist went through what, at the time, readers probably experienced as a surreal change, i.e., a gender transformation.

Shakespeare's plays and the Bible continue to offer a font for artists to draw on to describe their own inner states and observations of the world. They help us iterate through aesthetic insights.

Think also of Uncle Tom's Cabin. Not only did it offer a tangible narrative of the horrors of slavery, its lead character would be raised in a later era to symbolize an overly pliant Black American, too eager to endure racist slights to curry favor with their oppressor.

What impact did this have? This created several elements that allowed the increasingly democratic literati to join a dynamic frame of cultural references, putting personal and social events in literary contexts using allusions and alliteration. Moreover, it supplied both psychological and intellectual stimulation and overall cultural alignment, i.e., depth in a particular novel and breadth.

Improvisational Improvements

In addition to novels, even Western music evolved from its Pythagorean reductive ails. What Ted Gioia calls the Africanization of music. The inclusion of the African diaspora represented the single most consequential impact on western music and cultural history writ large. This is blues, jazz, country(!), and ultimately rock and roll and hip hop. Music was not the staid performance of analytically driven composers but the spontaneous expression of artists and commoners experiencing the daily psychological struggles and pains of artists and commoners.

This improvisational style and its orientation towards the process rather than product perhaps inspired movements in other modes of art. Improv comedy, of course, and the work of visual artists such as Keith Haring intentionally draw audience attention to how great works are created. We do not just marvel at works of art and interpret them at a distance; we are engulfed and, in an important way, become part of the work itself.

In other words, we have the answers that we need. We need to ensure that the robots build better humans, and not just that the humans continue to build better robots.

What is art? Why is humanly created art interesting?

We are not in the business of seeing art for art's sake. There is no objective orientation for the beauty of art that was created using tools that do not speak to a deep, i.e., a conscious being or soul. Art is interesting because it tells the artist's story. It reveals the depth of one's soul. It's what Neal Stephenson calls the microdecisions of the artist.

Is it cool for a car to drive faster than a human? Useful, yes, but the Olympic Games don't involve a guy in a Hyundai driving past a runner. We are interested in human capacity and not those of a soulless machine. We can talk again when machines do gain consciousness, but even then the interest of humans will largely be in other humans. We understand ourselves through the shared reality we inhabit.

Misalignment of the tech world

What's the problem? Can we trust tech companies to do the right thing? Can we trust them to build tools that advance human flourishing through care and devotion to holding not only their own values and priorities but the welfare of their consumers' longer-term interests? Can we trust them to align individual and social interest in a way that allows everyone to flourish?

Probably not.

I'm convinced that if Steve Jobs were alive today he would bemoan his own success with the smartphone. The device symbolizes the flawed impulses and direction of Silicon Valley's use of technology and what we can call "versioning of consumer capitalism," i.e., the idea that companies can serve their own needs over the welfare of their consumers and society writ large.

Neuroscientist Andrew Huberman and social psychologist Jonathan Haidt have shown that smartphones narrow our attention into a spotlight. Instead of immersion in the broad periphery of life, the screen forces us into tunnel vision, a mode of vigilance that heightens anxiety. Empirical studies confirm this narrowing, see here.

Linda Stone calls it the continuous partial attention phenomenon. The endless flicking of notifications keeps us connected to everything and nothing. It is shallow alertness, a state that corrodes focus and exhausts the nervous system.

Picture a modern aging and retired man, trying to spend time with their granddaughter but driven to distraction as their days have been filled with screen time. Not on the wide displays that promote peripheral vision but the narrow ones that, studies show, put us in heightened states of alertness and anxiety through the spotlight attention they foster.

Next Steps

We can do better. LLMs can now get an app started. While its important to qualify their knowledge as "flatland" in nature, they are an amazing piece of technology that represents a culmination of our digital reality, one humans have crafted not just through our time and attention over the past several years, each time we posted our thoughts and feelings on Reddit and Instagram, but also the very cumulative literal toil of hundreds of thousands of years of human civilization. Every piece of writing, ancient and modern, has been stripped of its intellectual property and funneled into these systems.

Speaking of which, we should spend some time staring at the twisted logic of judges that prevent copyright laws from preventing this unprecedented IP theft through a tortured interpretation of fair use.

Anyway, we are where we are. We can build and iterate on tools based on ideas without a dedicated scrum team of designers, full-stack devs and QA testers. While it's still wise to hire talent who understands tech and can do better than deploy vibe-coded idea apps to the wild, you can always get started. Demonstrate the capacity of tools to serve individuals and not the apps and screens that have driven us mad.

The largest benefit in education is a tutor. Someone who can serve as a personal guide as well as implicit inspiration towards bettering ourselves, correcting our errors and leading us to novel fields and levels of knowledge. We all have that now. We can become the type of person, with the type of skillsets that are not just useful for someone else but can serve the original function of mass education of any sort: to build better citizens, people, religious or moral characters. We can ride the ups and downs of a typical life, and hopefully better prepare for the ups and downs of an increasingly frightening coexistence with sentient and powerful robots.

Better Tools

For my own part I am:

1. Pursuing reserach on building AGI with Morality

2. Working on authentic self-expression with LLMs as a tutor not a substitute

3. Building tools that align tech with human interest.

Prescriptions: What we want

• - Tutors that help us master forms of self expression in art

• - Guidance for learning new subjects by optimizing things like memory, recall, and application

• - Social Networks that find people of like interests

• - Messaging apps that prompt us to message friends/family members at "healthy" intervals according to closeness, last message time, etc.

• - Work chat apps that auto-prioritize messages

Proscriptions: What we don't want

• - Applications that reduce human agency to pattern recognition

• - Tools that optimize for engagement over well-being

• - Systems that fragment attention rather than deepen focus

• - Interfaces that narrow rather than expand our cognitive periphery

• - Technology that employ John Doer’s OKRs like YouTube did originally, pushing political content to trigger us into binge watching viral reels of Jordan Peterson.

When literacy reshaped our souls we responded with literature. Let us ensure that this new intelligence does not leave our souls impoverished. Let us ensure that it deepens them.