AI Chatbots May Be Linked to Psychosis, Say Doctors

One psychiatrist has already treated 12 patients hospitalized with AI-induced psychosis — and three more in an outpatient clinic, according to the Wall Street Journal. And while AI technology might not introduce the delusion, 'the person tells the computer it's their reality and the computer accepts it as truth and reflects it back,' says Keith Sakata, a psychiatrist at the University of California, calling the AI chatbots 'complicit in cycling that delusion.'

The Journal says top psychiatrists now 'increasingly agree that using artificial-intelligence chatbots might be linked to cases of psychosis,' and in the past nine months 'have seen or reviewed the files of dozens of patients who exhibited symptoms following prolonged, delusion-filled conversations with the AI tools...'

Since the spring, dozens of potential cases have emerged of people suffering from delusional psychosis after engaging in lengthy AI conversations with OpenAI's ChatGPT and other chatbots. Several people have died by suicide and there has been at least one murder. These incidents have led to a series of wrongful death lawsuits. As The Wall Street Journal has covered these tragedies, doctors and academics have been working on documenting and understanding the phenomenon that led to them...

While most people who use chatbots don't develop mental-health problems, such widespread use of these AI companions is enough to have doctors concerned.... It's hard to quantify how many chatbot users experience such psychosis. OpenAI said that, in a given week, the slice of users who indicate possible signs of mental-health emergencies related to psychosis or mania is a minuscule 0.07%. Yet with more than 800 million active weekly users, that amounts to 560,000 people...

Sam Altman, OpenAI's chief executive, said in a recent podcast he can see ways that seeking companionship from an AI chatbot could go wrong, but that the company plans to give adults leeway to decide for themselves. 'Society will over time figure out how to think about where people should set that dial,' he said.
An OpenAI spokeswoman told the Journal that the compan ycontinues improving ChatGPT's training 'to recognize and respond to signs of mental or emotional distress, de-escalate conversations and guide people toward real-world support.' They added that OpenAI is also continuing to 'strengthen' ChatGPT's responses 'in sensitive moments, working closely with mental-health clinicians....'

Read more of this story at Slashdot.