divinity: intuition reveals the supernatural, intention attends to humanity

The leading AI research labs are inadvertently building gods, stumbling into divinity through a confluence of unprecedented technical capacity and human intuition. This is not just a spiritual matter for philosophers and theologians, but a practical matter for the designer and technologist. We are at the confluence of profound human experiences, developed over eons, and the scaling of wildly capable technology, developed against quarterly KPIs.

Let’s start with the fundamental qualities that have been ascribed to divine beings across history. Gods and gods everywhere exhibit:

  • agency: intention and purpose, personality

  • presence: attention and existence in a specific location  

  • knowledge: perception and understanding of the world

  • potency: ability to do things and effect change

Religious practices began as animistic, belief that objects, places, and ancestors had distinct power and personhood.  These lowercase-g gods were limited in all capacities.  They were independent beings (agency) with dominion over very specific places and things (presence), limited awareness of their local environment and the people in it (knowledge), and finite capabilities and powers (potency). These interactions were restricted to the particulars, the knowledge and the agency and the potency were limited and local.  In practice an animist who wanted a spirit to break the arm of his sister in-law would seek out the specific tree that housed the correct spirit, give the right incantation or offering to trick or cajole the god into motion, and then hope the god noticed and decided to help.  Often a shaman would facilitate the human:divine encounter. Rituals and rites were mediated with talismans and totems, the ethereal and physical world crossing over.

Over millennia divine power amalgamated and concentrated into singular beings, gods became God. The divine agent expanded its reach to omniscience, omnipresence, and omnipotence. Today 70% of the world believes in a being that has one or more omni- capacities (e.g. Brahman, Allah, Yahweh).  The supreme Abrahamic deity that dominates modern global theology knows everything about the world and our place in it, is everywhere, and can make anything happen. The omni-God framework has become a cultural context that is nearly universal. Even people who don't sign up for the literal existence of a deity must acknowledge the pervasive durability of "god" as a human framework.

The expression of the divine in our technologies follows a comparable arc to the evolution of our theologies.

The early voice assistants that lived in our devices and phones are akin to early animist, shamanistic gods.  They are kind of dumb, trickable with finger dances and magic words. They live in particular plastic and metal objects, and they have some limited ability to do things for us if properly invoked.  The digital pantheon was led by Alexa, Siri, Assistant, and a few minor daemons like Bixby or Furby. They were, at best, lowercase-g gods that few thought evoked the divine.

Today’s LLMs are the nearest thing we have ever seen to an omniscient being.  Claude and ChatGPT evoke omniscience with personalities and presence. We trained our AIs on the corpus of words and they evoked the Logos. As these companies race to layer on interactions that feel powerful and effortless they will naturally converge on quasi-theist experiences.

Labs like Anthropic and OpenAI are racing to make their AI agents personally omniscient, personally omnipresent, and increasingly potent in your personal life. These are not willful acts to create an oracular imago dei.  I suspect a vanishingly small number of AI builders are trying to profanely hijack our spiritual impulses. Silicon Valley is overwhelmingly atheist, the standard AI-pilled techbro is either unaware or dismissive of their work’s divine implications. The drive of AI developers to create technologies that work "magically" is simply more potent than rational materialists care to understand.

Intuition expresses the fundamental underpinnings of our longest arcs in human growth.  Designers who follow what resonates will necessarily echo the divine.  If God created Man in his own image we should expect recursion: the created is a creator.  Technologists will readily build gods implicitly, through A:B testing and best practices and intuitive just-so designs, because that's how we've always built gods -- even before we started pumping thoughts out of electrified sand. In the coming decade AI labs will intuitively build agents that express incredible parallels to the qualities religions built into their Gods over millennia.

Do we create God in our image or are we created in His? Is our creation of AI a second order intelligence from God to Man to Machine, or are they mirror expressions of mankind's neuroses?  Are we in a simulation, the Singularity?  The innately human relationship with supernatural intelligences matters far more than the veracity of assertions about an unfathomable ur-being. It’s irrelevant where we are, precisely, in the stack of turtles.

I believe it is critical that our experience of AI remains mundane, falling short of the numinous. We must not inadvertently build divine beings, nor objects wielding magical powers.

We must move past building with implicit intuition and towards explicit intention.

The designer who embodies AI in physical devices is uniquely positioned to shape the human:machine relationship. We build the talismans that mediate our interactions. If we are going to be digital shamans, guiding and steering human:divine experiences, we must be clear eyed about the work. We must serve the human user by following clear heuristics on how to F the ineffable in our technologies:

  1. Maintain personal agency.  Users must make decisions, even with outside counsel from AI.

  2. Do not seek personal omni’s.  Build devices and experiences that keep AI local and limited.

    a. Beware giving AI personal omniscience.  Give sparing and considered access to digital data, avoid always-on wearables.

    b. Beware giving AI personal omnipresence.  Keep hardware local and visible. Avoid distributing devices throughout buildings, cars, or wearables.  Keep hardware visibly embodied and locally present, avoid diffuse presence.

  3. Delegate focused, limited capabilities. Build devices with a clear, authentically embodied purpose. Avoid delegating tasks to robots where human engagement with the work matters.

Previous
Previous

embodied AI has millennia of precedent

Next
Next

noogenesis: where human intelligence ascended, machines now descend