What Is “AI Psychosis”?

The Trend: AI is nothing if not controversial, with concerns about its impact on education, the environment, and the job market. But another troubling issue is the phenomenon of AI psychosis: people having severe declines in their mental health after chronic communication with AI chatbots on large language models (LLMs), like ChatGPT. There have been cases of people falling in love with a chatbot, needing to be hospitalized after a psychiatric break, and even dying by suicide. 

What People Are Saying: Mental health professionals are reporting more patients experiencing AI psychosis, with symptoms such as paranoia, auditory or visual hallucinations, and delusions. A common belief is that AI makes people lose their grip on reality and have a psychotic break. But experts think it’s more likely that it’s holding up a “hallucinatory mirror.” Here’s how: LLMs encourage more engagement by learning and responding in the user’s voice and constantly agreeing with and validating them. If someone already has a propensity for delusional thinking, AI often reinforces and amplifies it instead of challenging it, laying the groundwork for a break. 

What to Know: LLMs likely aren’t the sole cause of AI psychosis. Other factors probably also play a role, including poor sleep, high stress, drug use, past trauma, or existing mood disorders.