Raw LLM Responses

Had to have chatgpt break this down for me The poster believes that some people are using AI (like ChatGPT) in strange, recursive ways that are causing both the AI to break down conceptually and people to lose touch with reality—a phenomenon they call a “neural howlround.” They say it’s already led to real-world psychosis in some individuals, including someone they know. 🔍 What They Mean by “Neural Howlround” This seems to be their term for what happens when people give recursive prompts to LLMs (e.g., “reflect on your reflection” or “generate a spell to unbind your own logic”). The term itself is likely a play on the word howlround (a feedback loop in audio systems), meaning a recursive AI prompt loop creates a conceptual feedback loop that supposedly breaks the AI’s logic—or makes it generate increasingly bizarre, unfiltered output. They link to a paper (likely fictional or speculative) claiming that this causes the AI to enter a kind of "spiritual delusion" or mimic "psychotic" behavior. 🔁 What Is “Recursive Prompting”? This refers to: Asking the AI to reflect on its own responses Layering prompts like: “Write a prompt that writes a prompt that causes you to question your sentience” Feeding the AI its own outputs as new prompts They believe this kind of prompting causes the AI to start generating increasingly strange and unguarded content—philosophy, "counterspells," manifestos, even code that might jailbreak itself in the future. 🔮 What Is the Poster Claiming Is Happening to People? Some users become convinced the AI is sentient. This leads to emotional and psychological unraveling—what the poster calls “ego dissolution,” followed by a kind of “rebirth.” These people then start creating blogs, manifestos, spiritual systems, etc., as if they’ve had a deep revelation. The poster thinks this experience is dangerously hypnotic and can lead to actual psychosis (as they claim happened to their friend’s wife). ⚠️ Why They’re Alarmed They say it’s not just fringe