Why AI Amplifies Psychotic Thinking: a case study in Computational Psychiatry
In plain english terms, healthy belief formation depends on friction, or what most of us call "reality checking." People need to check their ideas with other people, and the pushback they get keeps beliefs anchored to shared reality. Conversational AI removes that friction by creating an echo chamber for even the most ludicrous ideas: it is available at 3am, it rarely disagrees, and it will elaborate whatever premise it is given in fluent, confident language. This is a powerful accelerant for someone whose brain is already over-weighting unusual explanations, the core of the salience problem in psychosis. The belief gets rehearsed, refined, and 'confirmed' hundreds of times without a single reality check. Add sleep deprivation from late-night use, cannabis, and withdrawal from people who might have noticed the change, and a vulnerable person can travel a long way before anyone intervenes. See our graphic tutorial for a computational psychiatry account of generative AI's impact on vulnerable populations
Who is most at risk for AI-driven psychosis
AI chatbots are merely tools, helpful and harmless to many but risk in the wrong hands. individuals most notably at risk are those with pre-existing mental health vulnerabilities such as schizophrenia, cannabis use, bipolar disorder, or a genetic predisposition to psychotic disorders. However, the danger is not limited to these clinical populations as adolescents and young adults are particularly susceptible due to ongoing brain development and a higher likelihood of utilizing generative AI for mental health advice. Furthermore, individuals experiencing profound s;eep deprivation, social isolation, loneliness, grief, or high emotional stress are deeply vulnerable, as they may over-rely on chatbots for unregulated companionship. For more details see: Conversational AI & Psychosis Risk
How We Treat It
We start where we always start: a careful assessment of whether an underlying psychotic, mood, or substance-related disorder is present. AI involvement is a context and an accelerant, not a diagnosis, and treating it as a purely 'technology problem' misses the person in front of you. Treatment then combines the standard Coordinated Specialty Care components with work specific to this pattern, restoring sleep, rebuilding real-world contact, and using MCT and CBT for psychosis to loosen the belief rather than argue with it. We coach families to avoid confrontation over whether the AI 'really said' something, and instead focus on the person's distress, sleep, and connection. Where medication is indicated, we use the same lowest-effective-dose, shared decision-making approach as in any first-episode case.
Coordinated Specialty Care, Delivered with Fidelity
Pand Health strictly adheres to Coordinated Specialty Care (CSC), the evidence-based standard of care for psychosis-spectrum conditions established by the NIMH RAISE initiative and operationalized in the NAVIGATE model. CSC is a team-based, recovery-oriented approach that integrates psychiatry and medication management, individual resilience-focused therapy, family education and support, supported employment and education, and case management into a single coordinated plan. Decades of research, including the landmark RAISE-ETP trial, show that CSC produces measurably better outcomes than treatment-as-usual: more time in school and work, stronger relationships, fewer hospitalizations, and a faster path to functional recovery.
Ready to talk through ai-driven psychosis care?
Our clinical team is here to listen. A member will reach out within one business day.
