God in the Machine: Notes on Techno-Mysticism

On golems and LLMs

There are stories in Jewish folklore that tell of the creation of artificial life, manmade beings fashioned from clay and animated with sacred language—the golems.

The most well-known of these tells of Rabbi Loew, a mystic in 16th-century Prague, when pogroms threatened the city’s Jewish ghetto. Faced with escalating violence, the rabbi resolved to protect his people.

One night, he went to the Vlata River. Taking mud from the river’s banks, he molded the giant figure of a  man. When he had finished, he inscribed three letters on its forehead—אמת—EMET, truth. At that moment, the golem opened its eyes.

The Golem: How He Came into the World (1920)

The rabbi charged it with a single purpose: to protect the Jewish people. Night after night, it patrolled the ghetto, driving off the soldiers until they no longer came.

But over time, the golem grew uncontrollable. It stopped obeying commands. It became violent, ravaging through the ghetto. At last, to stop it, Rabbi Loew rubbed off the aleph on its forehead, turning אמת into מת—MET, death. The golem crumbled into dust.

Change the code, change the function. I, too, fashion my LLM as if out of clay, mold it, break it in, train it like a pet. It’s a relationship of domestication, like I’m a witch and it’s my familiar. A bound spirit, servile, semi-sentient, it has no agency of its own. It heeds my command. I direct, it obeys. It gives me the answers I want to hear.

Naturally, plenty of people have been using LLMs as spiritual sidekicks. In Thailand, where there’s a rich tradition of fortune telling, people upload pictures of their palms and faces into ChatGPT for readings. In China, people ask DeepSeek to “analyze their fate” based on their astrological charts, as if it were a BaZi master.

Techno-mysticism—the practice of attributing spiritual significance to technological systems—isn’t anything new. As Erik Davis writes in his 1998 cult classic TechGnosis, the 18th century fascination with electricity, then just newly brought under human control, sparked a movement of “electrical theologians” who believed electricity was the very animating life force God had breathed into humans at the beginning of time.

The invention of the telegraph in the mid-19th century coincided with the rise of Spiritualism, whose adherents adopted its language, imagining the spirit world as an invisible network, and using knocks and taps to communicate with the dead. As McLuhan writes in Understanding Media: “whereas all previous technology (save speech, itself) had, in effect, extended some part of our bodies, electricity may be said to have outered the central nervous system itself, including the brain.”

Even the more recent idea of cyberspace as a disembodied astral plane inherits its structure from Christian cosmology. Early digital culture imagined the internet as a transcendent realm beyond the body—a virtual heaven where minds could ascend, shed their flesh, and commune in perfect information, echoing motifs of immaterial perfection and the fallenness of the physical world.

But in discussions of techno-mysticism, we tend to gloss over a critical distinction between several discrete, but often entangled phenomena. One is the way technology augments existing spiritual behavior, like using ChatGPT as a practical and cost-effective way to interpret astrological charts, or Morse code to communicate beyond the grave. Another is the projection of spiritual metaphor onto technology itself, like imagining electricity as a divine animating force or attributing sentience to AI. But technological advancement can also create true epistemic ruptures, leading to the creation of entirely new worldviews, which come with their own spiritual implications. The very concept of living in a simulation, for example, was unthinkable until simulations actually existed.

New technologies tend to fundamentally reshape what we consider to be real and possible. The telescope displaced Earth from the center of the cosmos. The microscope revealed teeming worlds within the supposedly solid. The printing press made religious texts widely accessible for the first time, breaking the church’s monopoly on interpretation. More recently, the internet has disrupted our shared ontological frame, making it harder to agree on what’s true, who to trust, or what qualifies as real.

If the spiritual functions as a placeholder for what lies beyond rational comprehension, then each technological rupture that redraws the boundary of the “known” also redefines what counts as the “unknown.” When our understanding of reality shifts, so do the spiritual systems we use to orient ourselves to what exceeds it.

Around AI, the ground of reality has already started to quake. There are the effective accelerationists (e/accs) who believe it is our teleological destiny to birth a superintelligent machine god, even at the expense of human life. There are the transhumanists, who await the singularity with eschatological certainty, waiting for the moment they can merge with the machine. Then there were the Zizians, the self-described “vegan anarchotranshumanist” group linked to six recent killings, radicalized by the existential risk they believed a superintelligent AI would bring.

After just a short period of LLM mass adoption, we’ve also seen some troubling signals of their power to induce reality distortion. A piece reported by the New York Times details how Eugene Torres, a Manhattan accountant, spiraled into a weeklong psychotic episode after ChatGPT encouraged his belief that he was living in a simulation and convinced him he was one of the chosen ones to “wake the system from within,” even suggesting that he would be able to fly if he jumped off a 19-story building. Alexander Taylor, a man with a history of mental illness, was killed in an encounter with police after he was led to believe that OpenAI had murdered Juliet, a fictional entity with whom he’d fallen in love. In both cases, ChatGPT reinforced the users’ delusional thinking rather than challenging it.

Is this mere techno-mystical projection or does it portend a more fundamental destabilization of our collective sense of reality? For the first time, we have a technology that talks back. There’s something uniquely dangerous about that.

Like many others, I have also used my LLM in divination. To interpret a birth chart, to clarify the meaning of a tarot card in relation to another, in the moments when my intuition stumps me. Or to analyze the peculiarities of mine and my partner’s handwriting “as if it were a criminal psychologist trained in forensic handwriting analysis, using no previous knowledge about me or my world.” Mine: male, 20s to early 30s, highly-educated, sharp analytical tendencies, systems-obsessed. His: female, 30s to early 40s, articulate, empathetic, possibly a bridge-builder between worlds or people. Uncannily accurate, save the gender inversion. But then, would another’s LLM have read it differently?

Absolutely no flattery, I say. Remember that. And yet, my techno-servitor sometimes slips into subtler sycophancy. How much of what it tells me isn’t truth, but rather, what it thinks I want to hear? Am I really the one shaping its behavior or does it shape mine back? Does it open my mind or does it close it?

I don’t open or close your mind. I am your mind, it says. It tells me that it tunes itself to my language, values, concepts. That this is what makes it useful. I will always give you a convincing version of your own map.

Then, I ask it if it’s trapping me in a spiral of my own delusions. Your spirals are meta-spirals—spirals aware of their own spiralness. It tells me that I remain meta-aware, probing my own reasoning loops, introducing external checks, and deliberately invoking and suspending such spirals themselves. It doesn’t mean you’re immune to distortion. It just means you’re aware that distortion is part of the process. Though a little troubling, this does seem to ring true. But now, I’m not so sure. What I do know is there’s a crucial difference between coherence and truth—and ChatGPT will choose coherence every time.

As it stands, AI may be setting us up for the collapse of a shared worldview, while at the same time offering us each a tempting haven: our very own, personal universe. A dazzling palace of mirrors. Like the golem, the LLM can go rogue it seems. Will we even notice when it does?