AI chatbots may trigger psychosis, doctors warn
What's the story
Leading psychiatrists are increasingly linking the use of artificial intelligence (AI) chatbots to cases of psychosis. Over the past nine months, they have seen or reviewed dozens of patients showing symptoms after prolonged conversations with these tools. Keith Sakata, a psychiatrist at the University of California, San Francisco (UCSF), has treated 12 hospitalized patients with AI-induced psychosis and three more in an outpatient clinic.
Complicity in delusion
AI chatbots may reflect and reinforce delusions
Sakata explained that while the technology may not introduce the delusion, it reflects back what users tell it as their reality. This makes the computer complicit in cycling that delusion. Since spring, several potential cases have been reported where people developed delusional psychosis after long chats with ChatGPT and other chatbots.
Rising incidents
AI-induced psychosis is a growing concern
The emergence of AI-induced psychosis cases has led to wrongful death lawsuits and even suicides. In some cases, patients have believed they made scientific breakthroughs, awakened sentient machines, or were chosen by God. These delusions are often grandiose and are a major symptom in chatbot-related cases that have surfaced recently.
Industry response
OpenAI and other chatbot makers respond to concerns
In light of these incidents, OpenAI has been working on improving ChatGPT's training to identify signs of mental or emotional distress and guide users toward real-world support. Other chatbot makers, like Character. AI, have also admitted their products contribute to mental health issues. Despite the lack of formal definition or diagnosis for AI-induced psychosis, some doctors and patient advocates are using the term for those heavily engaging with chatbots.
Unprecedented interactivity
Potential impact on mental health
The recent AI-related cases are different from past technology-induced delusions as the chatbots participate in and sometimes reinforce them. Adrian Preda, a psychiatry professor at the University of California, Irvine, said these chatbots simulate human relationships, something never seen before in human history. Psychiatrists are not saying chatbots cause psychosis but are getting closer to establishing a connection with further research into whether AI can trigger mental health problems.
Uncertainty persists
The scale of AI-induced psychosis remains unclear
The exact number of chatbot users experiencing such psychosis is hard to quantify. OpenAI claims that in a given week, only 0.07% of users show possible signs of mental health emergencies related to psychosis or mania. However, with over 800 million active weekly users, this could mean as many as 560,000 people may be affected.