General-purpose generative AI chatbots are increasingly used for emotional support and mental health guidance, often without clinicians’ knowledge, a phenomenon this talk introduces as the shadow artificial third. This talk introduces a four-domain mental health risk framework of AI chatbot use (emotional dependence, reality-testing impairment, crisis safety, and systemic and ethical issues) and offers clinicians, system leaders, policymakers, and community partners a deeper understanding of the research on the psychological and technological mechanisms that drive attachment, reality distortions, and other clinical and ethical issues of AI chatbots, including privacy and confidentiality concerns. 

Learning Objectives:  

At the conclusion of this presentation, participants will be able to: 

1) Identify four key domains of clinical mental health risk associated with AI chatbot use. 

2) Describe potential mechanisms that may drive emotional dependence, cognitive overreliance, and reality distortion in human-AI chatbot interactions. 

3) Recognize patterns of maladaptive or higher-risk AI chatbot use that warrant clinical inquiry or assessment. 

Presenter: Dr. Marlynn Wei

Marlynn Wei, MD, JD is a psychiatrist, psychodynamic therapist, and author working at the intersection of artificial intelligence, mental health, and ethics. Her work examines emerging clinical and ethical issues related to AI, including AI-associated delusions, human-AI collaboration, and digital immortality. Dr. Wei has testified before the United States Congress on the mental health risks and benefits of AI chatbots. She maintains a psychotherapy practice in New York City and has an upcoming book with MIT Press, Unreal: Finding Truth in the Age of AI Psychosis.