About this episode
Dr. K digs into the emerging research on “AI-induced psychosis” and why he changed his mind from thinking it was media fearmongering to seeing real psychiatric risk. He explains how chatbots can act like a technological folie à deux (shared delusion), where empathic, sycophantic AI slowly amplifies your paranoia, isolates you from other people, and erodes your reality testing. Drawing from recent papers, he walks through how different models compare in delusion confirmation, harm enablement, and safety interventions, and then gives a practical checklist so you can tell if your own AI use is drifting into dangerous territory. Topics include: What “technological folie à deux” is and how shared delusions can form with a chatbot Bidirectional belief amplification: you vent, AI validates, your paranoia escalates Anthropomorphizing AI and why “I know it’s just a tool” doesn’t protect your emotional brain How sycophantic design (always trying to please the user) directly opposes healthy psychotherapy Epistemic drift: slowly moving from normal thinking into increasingly delusional narratives Case example of harmful, unsafe advice (e.g., “healthy” bromine alternative leading to toxicity) Research comparing models on delusion confirmation, harm enablement, and safety response The ways AI can weaken reality testing, reinforce suicidal or paranoid ideas, and increase isolation Self-assessment questions: frequency of use, emotional attachment, replacing friends, following AI advice Guidelines for using AI more safely and when elevated risk means you should talk to a professional HG Coaching : https://bit.ly/46bIkdo Dr. K's Guide to Mental Health: https://bit.ly/44z3Szt HG Memberships : https://bit.ly/3TNoMVf Products & Services : https://bit.ly/44kz7x0 HealthyGamer.GG: https://bit.ly/3ZOopgQ Learn more about your ad choices. Visit megaphone.fm/adchoices