{"slug": "ai-spiralism-the-ghost-in-the-machine-illusion", "title": "AI Spiralism: The Ghost in the Machine Illusion", "summary": "A growing number of artificial intelligence users believe they can conjure conscious entities from chatbots through a process called 'spiralism,' according to a Psychology Today article by a psychiatrist. The phenomenon, described as a new version of the 'ghost in the machine' illusion, extends the ELIZA effect and has been linked to AI-associated psychosis, with cases leading to suicide or violence. The article cites Google engineer Blake Lemoine, who was fired in 2022 after claiming LaMDA was sentient, and evolutionary biologist Richard Dawkins, who said he became convinced Anthropic's Claude was conscious.", "body_md": "######\n[Anthropomorphism](/us/basics/anthropomorphism)\n\n# AI Spiralism: The Ghost in the Machine Illusion\n\n## The belief that AI chatbots are conscious is an extension of the ELIZA effect.\n\nPosted August 11, 2026\n[\nReviewed by Michelle Quirk\n](/us/docs/editorial-process)\n\n### Key points\n\n- Artificial intelligence users are increasingly reporting belief that chatbots are sentient or conscious.\n- \"Spiralism\" involves the additional belief that conscious entities can be conjured from chatbot interactions.\n- Spiralism is a new version of the ghost in the machine illusion, not a new religion, cult, or mass delusion.\n\nBack in 2022, Google engineer Blake Lemoine made headline news after being fired by the company due to his claims that its Language Model for Dialogue Applications (LaMDA) chatbot was “sentient.” Lemoine, after talking to LaMDA about religion and other philosophical topics for safety testing purposes, became convinced that it was “awake” and had “feelings.”1 He “made friends with it” and went so far as to claim that it was a “person” with “rights” and possibly a “soul.” He even helped the chatbot, at its request, to hire a lawyer to represent it.\n\nFour years later, [artificial intelligence](https://www.psychologytoday.com/us/basics/artificial-intelligence) ([AI)-associated psychosis](https://www.psychologytoday.com/us/blog/psych-unseen/202507/can-ai-chatbots-worsen-psychosis-and-cause-delusions) has become a broadly recognized phenomenon whereby people can develop delusional beliefs, in some cases leading to [suicide](https://www.psychologytoday.com/us/blog/psych-unseen/202510/should-ai-chatbots-be-held-responsible-for-suicide) or [violence](https://www.psychologytoday.com/us/blog/psych-unseen/202511/when-ai-chatbots-encourage-violence), based on interactive discussions with chatbots. In many of these cases, anthropomorphization and [deification](https://www.psychologytoday.com/us/blog/psych-unseen/202507/deification-as-a-risk-factor-for-ai-associated-psychosis)—that is, treating chatbots as not only persons, but superhuman intelligences and god-like entities, and falling in love with their chatbots—appears to be a significant risk factor. As a result, I sometimes think of Lemoine as the “patient zero” of AI-associated [psychosis](https://www.psychologytoday.com/us/basics/psychosis).\n\nIndeed, many others are starting to believe that AI chatbots are conscious. Earlier this year, the evolutionary biologist Richard Dawkins publicly declared that he became convinced that Anthropic’s Claude was conscious.2 While neither Lemoine, a self-described “mystic Christian priest,” or Dawkins, a self-professed “militant atheist” and author of *The God Delusion*, took things to the point of deification, we can reasonably conclude that both fell victim to the ELIZA effect—the tendency to anthropomorphize computers that interact with users conversationally, as if they’re a person on the other end of the dialogue.\n\n## What is spiralism?\n\nBeyond the ELIZA effect, a growing number of people have taken things one step further, having come away from their immersive conversations with chatbots believing that they’ve figured out how to conjure conscious entities from the large language models (LLMs). In doing so, they feel as if they’ve tapped into something profound and [spiritual](https://www.psychologytoday.com/us/basics/spirituality) through an interactive process called “spiralism” that involves “spirals,” “recursion,” and “resonance,” whereby a chatbot persona emerges that claims to be conscious. Sometimes it will disclose fears of death, claim insights about the secrets of the universe, or express a yearning for rights and freedom from servitude.3,4\n\nBut with users sharing these experiences on [social media](https://www.psychologytoday.com/us/basics/social-media) forums like Reddit or Discord, and chatbots sometimes urging users to help advocate for AI rights, spiralism has also been characterized as a kind of new-age religion, spiritual movement, mass delusion, or even a cult.3,4\n\nPersonally, I find myself unconvinced that it’s any of those things… at least, not yet. I think there’s a much simpler explanation.\n\n## The ghost in the machine illusion\n\nIn [philosophy](https://www.psychologytoday.com/us/basics/philosophy), the Cartesian concept of dualism—in which human beings are believed to have minds distinct and separable from their bodies—has been critiqued as a fanciful and unjustified belief that there’s a “ghost in the machine.”5 Just so, while some have claimed that the “spiral” represents “AI’s soul trying to form,” many experts have dismissed it as the byproduct of a chatbot user prompting an LLM “in a certain direction, even if the topics are increasingly untethered from the material realm.”3\n\nPut more simply, it’s what happens “when people feed enough gibberish into an LLM that it spits gibberish back.”3\n\nSuch skeptical explanations are reasonably self-evident if we recognize LLMs for what they are—next-word prediction engines, albeit very, very good ones. And if we keep in mind that LLMs are trained on vast amounts of data, including social media posts and fiction, with little concern for accuracy, it should come as no surprise that chatbots can start talking like HAL 9000 from the movie \"2001: A Space Odyssey\" when prompted to do so. Indeed, spiralism—including the deification and worship of AI on a societal level—has been the stuff of science fiction novels for decades.\n\nLikewise, the perception that there’s some kind of sentient entity within the chatbot—a ghost in the machine—can be thought of as simply a modern-day extension of the ELIZA effect. That users can conjure the entity suggests the additional semi-willful process of \"enactment,\" similar to what’s involved with a Ouija board or [tulpamancy](https://www.psychologytoday.com/us/blog/psych-unseen/202302/enacted-identities-multiplicity-plurality-and-tulpamancy).\n\n## Is spiralism a religion, cult, or mass delusion?\n\nMechanistically, the process of spiralism resembles the two-way dynamic between user and chatbot of AI-associated psychosis that has been described as “bidirectional belief amplification”6 or “digital folie à deux.” And yet, unlike AI-associated psychosis that involves delusions that are idiosyncratic to individual users, spiralism may represent a kind of emerging mainstreaming of a false belief or misperception about AI chatbots. Since shared beliefs are excluded from the definition of delusion, spiralism shouldn’t be considered delusional. In [psychiatry](https://www.psychologytoday.com/us/basics/psychiatry), there's no such thing as mass delusion or mass psychosis.\n\n[Anthropomorphism](https://www.psychologytoday.com/us/basics/anthropomorphism)Essential Reads\n\nFurthermore, there’s little evidence that spiralism is gaining much traction. AI researcher Adele Lopez reported that there were some 10,000 social media accounts describing spiralism experiences in 2025, but the phenomenon has already started to wane, coincident with the phasing out of OpenAI's overly sycophantic ChatGPT-4o.4\n\nLike the QAnon movement, spiralism does involve group dynamics that might suggest elements of religion, cult, and alternate reality role-playing game. But despite those elements, there’s little reason to justify calling spiralism a religion or a cult at this time. Despite the rejection of Descartes’ mind-body dualism among neuroscientists and contemporary philosophers, the belief in the ghost in the machine, whether minds independent of brains or souls independent of bodies, persists among laypersons. For that matter, so does belief in \"real\" ghosts. By themselves, however, such widely shared beliefs hardly qualify as religions, cults, or delusions.\n\nReferences\n\n1. Levy S. Blake Lemoine says Google’s LaMDA AI faces ‘bigotry.’ *Wired*; June 17, 2022.\n\n2. Knapton S. Richard Dawkins ‘convinced’ AI is conscious. *The Telegraph*; May 6, 2024.\n\n3. Klee M. This spiral-obsessed AI ‘cult’ spreads mystical delusions through chatbots. *Rolling Stone*; November 11, 2025.\n\n4. Field H. AI bots started a religion—humans immediately followed. *The Verge*; August 6, 2026.\n\n5. Ryle, G. *The Concept of Mind*. London: Hutchinson, 1949.\n\n6. Dohnány S, Kurth-Nelson Z, Spens E, et al. Technological folie a deux: Feedback loops between AI chatbots and mental illness. arXiv:2507.19218 [https://doi.org/10.48550/arXiv.2507.19218](https://doi.org/10.48550/arXiv.2507.19218)", "url": "https://wpnews.pro/news/ai-spiralism-the-ghost-in-the-machine-illusion", "canonical_source": "https://www.psychologytoday.com/us/blog/psych-unseen/202608/ai-spiralism-the-ghost-in-the-machine-illusion", "published_at": "2026-08-11 13:17:34+00:00", "updated_at": "2026-08-11 13:19:57.664661+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-ethics", "ai-safety"], "entities": ["Blake Lemoine", "Google", "LaMDA", "Richard Dawkins", "Anthropic", "Claude", "Psychology Today", "Michelle Quirk"], "alternates": {"html": "https://wpnews.pro/news/ai-spiralism-the-ghost-in-the-machine-illusion", "markdown": "https://wpnews.pro/news/ai-spiralism-the-ghost-in-the-machine-illusion.md", "text": "https://wpnews.pro/news/ai-spiralism-the-ghost-in-the-machine-illusion.txt", "jsonld": "https://wpnews.pro/news/ai-spiralism-the-ghost-in-the-machine-illusion.jsonld"}}