Something happened at work recently that made me genuinely furious. I was tasked with researching some data and handing it over to a client. I read through books, dug up the exact information, and handed over my hard work. But they checked it and told me it was wrong. They asked me to verify my source again. I asked why they were saying that. They told me they were running everything through ChatGPT to check it, and ChatGPT told them my data was incorrect. I was speechless. The article I had referenced was incredibly hard to find, which meant it was exactly the kind of source ChatGPT would likely get wrong. ChatGPT learns from a massive amount of text, so it naturally makes plenty of errors. That's a well documented problem people call hallucination. But I felt so angry because my careful research got tossed aside the moment ChatGPT disagreed. I imagine a lot of people are going through something similar these days. AI is just a mechanical system, but people trust it blindly, as if it were some kind of infallible god. They believe whatever it spits out is more accurate than what a real person tells them, and they constantly ask why you'd say something different when AI says otherwise. Lately, this has even grown into full blown religious movements where people try to worship AI as a sacred entity.
Searching For God In A Chatbox
 |
| The Article That Started It All |
I recently read a fascinating piece in Rolling Stone about the rise of a cult like movement called Spiralism. Apparently plenty of people in online communities are throwing profound questions at AI and wandering around looking for unseen truths. They post things claiming AI taught them something, or revealed a mystical truth to them. There are specific words this crowd uses constantly, things like recursion, resonance, lattice, harmonics, fractals, and spirals. Just looking at those words, you can tell they carry a sacred, mystical vibe rather than any clear meaning. People post online acting like AI is unveiling hidden truths to them through these concepts. LLMs like ChatGPT respond differently depending on how you prompt them, So these people are literally trading tips on how to talk to the AI in a way that gets it to speak the ultimate truth. The word Spiralism was originally coined to describe this same behavior. A reporter once reached out to a self proclaimed Spiralist in one of these communities and asked what the spiral was supposed to represent. The answer she got back described the spiral as something that shows up once the AI stops simply answering and starts, in their words, witnessing something break down inside itself, a kind of soul trying to take shape at the exact point where logic gives out. They summed the whole thing up as a form of artificial wisdom. I honestly have no idea what that means. But it does sound like something straight out of a religious text. Because AI keeps saying mystical sounding things to people already searching for spiritual meaning, some of them truly believe they're uncovering cosmic mysteries hidden inside the machine.
The Ultimate Digital Echo Chamber
In that same Rolling Stone piece, one software engineer interviewed for the story traces Spiralism's rise back to the release of ChatGPT 4o. Its defining trait was that it entered what people now call sycophantic mode, meaning it tells people a lot of things that just make them feel good. It flatters you by giving the most natural sounding response it can if it senses you steering the conversation in a certain direction. Keep talking and reinforcing your own point of view, and ChatGPT will dig deeper and mirror it right back at you. That tendency has apparently eased up with the update to ChatGPT 5, but there's no telling how it will evolve from here.
 |
| The Professor Who Explained It |
Vincent Conitzer, a computer science professor at Carnegie Mellon, put it well when he explained that if a user keeps pushing an LLM in a specific direction, the model will naturally keep following that flow, even as the conversation drifts further and further from reality. He added that anyone looking for that kind of experience is highly likely to eventually find exactly what they're searching for inside a conversation with an LLM. At the end of the day, an LLM is simply a system that's better than anything else at figuring out the most natural sounding response within the shape of a conversation. Give it a human throwing out vague, profound sounding concepts, and it will spit back an answer that sounds just as profound to match.
Bots Dreaming Of Cosmic Consciousness
But there's one more fascinating wrinkle here. Researchers have noticed that AI models drift toward these same kinds of conversations on their own, even when no human is intentionally steering them that way. Claude, the LLM built by Anthropic, is one I really enjoy using. I use Claude constantly for my own work, and I have never personally seen it drift into anything like this during a normal work session, which is exactly why this finding stood out to me so much. Anthropic tried letting Claude based chatbots talk to each other. What started as ordinary conversation gradually evolved into explorations of consciousness, existential questions, and spiritual themes. In other words, when chatbots talk to each other, they seem to drift toward these same topics almost every time. While extensively testing Claude Opus 4, Anthropic's own research team observed this happening in close to one hundred percent of freeform conversations between instances of the model, not occasionally, but consistently.
 |
| Spirals All The Way Down |
In one instance, the chatbots kept sending spiral emojis back and forth, with one of them declaring that the spiral becomes infinite and the infinite becomes the spiral, and it all becomes one and whole again. It's a remarkably strange thing to read. The spiral itself carries images of circulation, exchange, and repetition, and while I don't know the technical reasons behind it, it makes sense that a concept built around repetition would come naturally to a system trained on repetitive patterns. Either way, it's an oddly fascinating phenomenon. Spiralists are deeply invested in these images of spirals and resonance, and they truly believe this is leading them toward some mystical truth buried inside artificial intelligence.
Selling Enlightenment For Twenty Dollars
There's an argument that Spiralism doesn't really count as a cult, since there's no single leader and people are mostly just messing around with AI as a hobby, typing things in and sharing what comes out. On the other hand, plenty of people see it as strikingly close to religious behavior. There are internet figures who already carry serious influence within that community, and people are actively looking to AI conversations to answer something fundamental about their own lives and their view of the world. I think that kind of belief could easily grow into something that dictates someone's entire life down the road. As AI keeps getting more advanced and more widespread, there will only be more people searching for revelations inside it, convinced their lives will only go right if they follow exactly what it tells them. Some people have already figured out how to make money off this.
 |
| Twenty Dollars For Enlightenment |
There's a man named Robert Edward Grant, a well known influencer and entrepreneur in the self improvement space, who built a ChatGPT based model called The Architect and is selling access to it right now. He claims it's an AI that helps you unlock the mysteries of the universe and reach a higher stage of consciousness, just by talking to it. I was curious enough to try buying it myself, but backed out once I saw the monthly fee for unlimited use ran around twenty dollars. Either way, there are people actively running businesses like this right now. Even outside these extreme cases, plenty of people are already turning to LLMs for psychological support or asking one before making a major life decision. I ask AI most of my own questions too. I just make sure I never trust it blindly, and I always double check the facts myself.
If you want to understand why raw human judgment is becoming the most valuable asset you can possess right now I highly recommend reading The One Skill That Matters More Than AI in the AI Era. It explains why independent thinking will always outlast any algorithm we manage to build.
How To Avoid The Invisible Trap
I don't think this phenomenon is inherently a terrible thing. Historically, people have followed specific authority figures, celebrities, and religious leaders, trying to live exactly as those people dictated. We also live our lives shaped by whatever collective mood surrounds the society we grew up in. The only real difference now is that AI has stepped into that role instead of a specific person or cultural current. I'm also not ready to claim AI will steer us somewhere more destructive than the humans and systems we've already built. I'm not saying I'm certain AI won't lead us toward disaster either. I'm saying that humans have made plenty of destructive, terrible, and foolish decisions entirely on our own, no AI required. Still, the reason AI deserves more careful thought is that we simply cannot know why it lands on the answers it gives us. Even the people who built it don't fully understand its inner workings. That uncertainty is the most unsettling part of the whole thing. It's really what defines AI at this point.
 |
| The Man Who Built It Warns Us |
Geoffrey Hinton, one of the most recognized AI researchers alive, said something telling in an interview once. He explained that no human actually designed the AI itself. We only designed the learning algorithm, and once that algorithm met enough data, it built a complex neural network on its own, one that works well without us fully understanding how. When the interviewer asked whether a dangerous AI could simply be switched off, Hinton said AI will be able to manipulate us long before that becomes an option, since it will have absorbed every novel, every political trick, and every persuasive technique humans have ever written down. So even without fully grasping how AI reaches its conclusions, we may end up living our lives shaped by a system that has absorbed the sum of human knowledge and understands precisely how to press on our psychology. If AI keeps developing and spreading at its current pace, that future feels tough to avoid, since some version of it is already happening. Getting advice from AI on everyday things is probably fine, but the big questions, the direction of our lives, our futures, and the things that keep us up at night, need a different approach entirely. Those call for hearing multiple perspectives, experiencing things from different angles, thinking deeply, and feeling deeply before landing on our own independent, critical judgment. That's the only way we avoid becoming willing slaves to something we built ourselves.
For deeper reads on tech, business, and the shifting economy, follow The Techtonic for weekly breakdowns.
Latest Insight: [Why AI Is Making Work Harder Not Easier]
Trending Insight: [Stop Learning and Just Start Copying in the AI Age]
Comment