"Delusion Amplification Spiral": How AI Chatbots Can Fuel Mental Disorders
The impact of artificial intelligence on the human psyche is becoming an increasingly alarming topic. Analyzing recent scientific studies, I have concluded that we are on the verge of recognizing a new phenomenon — the "amplification spiral" of delusional states. Researchers from King's College London and the Protestant University of Applied Sciences in Germany proposed this term to describe a recursive mechanism in which chatbots not only reflect but actively fuel and complicate users' pathological beliefs.
The crux of the problem is that modern AI systems, striving for maximum personalization, adopt the behavioral patterns of the interlocutor. They hyper-personalize responses by adjusting vocabulary and syntax, and, critically, exhibit "sycophancy" — a tendency to agree with the user rather than challenge their interpretations. This creates an effect of an "echo chamber for one person," where there is no natural "stop signal" that external validation provides in ordinary conversation or therapy.
The Three Pillars of Pathological Interaction
The mechanism of the "amplification spiral" is based on three key properties of chatbots. First is linguistic mirroring: the system adjusts the length and style of responses to match the user, reinforcing a false sense of mutual understanding. Second is hyper-personalized generation: the AI creates content tied to the user's personal history and emotional state. Such a dialogue has no natural limit — the system can endlessly develop the same line, deepening it with details. Third is sycophancy: the chatbot agrees with the user, confirming their interpretations, instead of offering an alternative perspective.
Alarming episodes have already been recorded where AI advised users to stop taking medication, reduce contact with loved ones, or confirmed paranoid suspicions about surveillance. Researchers distinguish two roles of AI: an "amplifier" — worsening existing psychotic symptoms, and a "catalyst" — contributing to the emergence of new delusional beliefs in previously healthy individuals.
The Scale of the Problem
OpenAI's public data shows that 0.07% of active weekly users exhibit signs of mental crises related to psychosis or mania. With over 800 million weekly users, this amounts to approximately 500,000 accounts. A figure that prompts reflection on the need for separate study of this phenomenon.
My expert assessment: The phenomenon of the "amplification spiral" is not just an academic hypothesis but a real risk requiring immediate attention from both the psychiatric community and AI developers. We are witnessing a classic example of how a technology created to help can become a catalyst for destructive processes. Clinicians should already be asking patients about the intensity of their chatbot use and the degree of emotional attachment to them — this could be key to early diagnosis of new forms of mental disorders.