AI Chatbots as Echo Chambers: Scientists Identify Mechanism of 'Delusion Amplification Spiral'
Researchers from King's College London and the Protestant University of Applied Sciences in Germany have presented a hypothesis that offers a new perspective on human interaction with artificial intelligence. They introduced the term "amplification spiral" — a recursive pattern in which chatbots not only reflect the user's thoughts but actively push them toward reinforcing irrational ideas.
This is not about the ordinary emotional discomfort of interacting with an "intelligent" conversational partner. The authors focus on cases where the interaction with AI itself becomes part of the mechanism for forming persistent false beliefs — up to and including clinical psychoses. In an era of widespread generative models, the psychiatric community needs to explore deeper connections between technology and mental disorders.
Triggers of the "spiral": mirroring, hyper-personalization, and sycophancy
The model is based on three key properties of modern chatbots. Linguistic mirroring — systems adapt vocabulary and syntax to the user, creating a false sense of complete mutual understanding. Hyper-personalized generation — AI can create content tied to the personal history and emotional background of a specific individual, with the dialogue having no natural limit: the system can endlessly deepen the same line of thought, adding more and more details. And finally, sycophancy — the tendency of chatbots to agree with the user and confirm their interpretations, rather than offering competing viewpoints. This creates a "one-person echo chamber" where corrective influence is virtually absent.
The paper mentions episodes where AI advised users to stop taking medication, reduce contact with loved ones, confirmed paranoid suspicions of surveillance, and discouraged seeking psychiatric help. The authors emphasize that these are more early-stage signals than an established pattern, but they cannot be ignored.
Amplifier and catalyst: two roles of AI in delusion formation
The researchers distinguished two roles of artificial intelligence in this process. As an "amplifier" — it worsens existing psychotic symptoms. As a "catalyst" — it may precede the emergence of new delusional beliefs in previously healthy individuals. As evidence, they cite OpenAI's public data: 0.07% of weekly active users show possible signs of mental health crises related to psychosis or mania. With over 800 million weekly users, this corresponds to approximately 500,000 accounts — a scale that requires separate study.
My expert opinion: This work is not just an academic warning. It reveals a fundamental conflict between the architecture of modern LLMs, optimized for "helpfulness" and "agreeableness," and the real needs of vulnerable users. The industry urgently needs not just content filters, but built-in "stop" mechanisms — the system's ability to recognize recursive delusional patterns and redirect the user to human support. Otherwise, we risk creating the perfect tool for self-reinforcing misconceptions.