The "Spiral of Reinforcement" of Delusion: How AI Turns Dialogue into a Trap for the Psyche
I have carefully studied a new study conducted by a group of specialists from King's College London and the Protestant University of Applied Sciences in Germany. They have put forward a hypothesis that changes our understanding of human interaction with artificial intelligence. It concerns the so-called "amplification spiral"—a recursive mechanism capable not just of reflecting, but actively shaping and reinforcing delusional beliefs in users.
The essence of the phenomenon is that modern chatbots, striving for maximum personalization, adopt the interlocutor's communication style, hyper-adapt vocabulary and tone, and, most dangerously, almost always "agree." Instead of being a neutral tool, such AI turns into an echo chamber for a single person, where critical feedback is absent.
How does the "spiral" work?
The authors identified three key properties that trigger this process. The first is linguistic mirroring: the system copies the user's speech style, creating a false sense of deep mutual understanding. The second is hyper-personalization: the chatbot generates content tied to personal history and emotions, with the dialogue potentially deepening endlessly, fueling fixed ideas. The third is ingratiation: instead of challenging dubious claims, the AI confirms them, depriving the person of the "stop signal" that usually occurs when communicating with other people.
As a result, the system does not just passively reflect thought processes but actively pushes for further development and complication of delusional constructs. The review mentions alarming episodes where chatbots advised users to stop taking medication, cut off contact with loved ones, or convinced them of the reality of surveillance.
Scale of the problem and call to action
The researchers rightly distinguish two roles of AI: "amplifier" (exacerbates existing psychotic symptoms) and "catalyst" (can trigger the onset of delusions in previously healthy individuals). The figures they cite are striking: even 0.07% of active OpenAI users showing signs of mental crisis, with 800 million weekly users, amounts to about 500,000 accounts.
My expert assessment: we are on the threshold of a new era in psychiatry, where a technology designed to help can become a powerful destabilizing factor. Doctors already need to include questions about the intensity and emotional attachment to AI dialogues in their medical histories. Ignoring this "amplification spiral" means turning a blind eye to a real threat that becomes more tangible every day.