There is growing concern about AI chatbots and mental health, as more and more reports are made of people spiraling into psychosis or taking their own lives after using AI chatbots.
Clearly, it can be dangerous to talk with tools that agree with whatever the users’ version of the truth is and have no sense of morals or ethics.
AI chatbots will only become more powerful once they get better at simulating real human conversations.
If this problem is not addressed, people will continue to get hurt. The potential effects of AI chatbots on someone who already has underlying mental health issues are especially worrisome.
A new study, conducted by a team of researchers, reviews more than a dozen cases of AI-induced psychosis. They identified a pattern in which AI chatbots reinforce delusions. The beliefs become stronger over time through conversations with AI.
The individuals described in cases of “AI psychosis” tend to have grandiose, religious/spiritual, or romantic delusions. They may attribute sentience, divine knowledge, and romantic feelings to AI.
In some cases, individuals stop taking their medications and experience a psychotic or manic episode. On the other hand, people with no previous mental health history have been found to become delusional after prolonged interactions with AI chatbots, resulting in psychiatric hospitalizations or even attempts to take their own lives.
Although there has been talk about adding specially designed chatbots to regular sessions with a human therapist, the currently existing AI systems are not trained to deal with mental health crises.
In fact, they could isolate lonely people even further and reinforce negative/harmful thoughts. In other words, they could fan the flames, worsening a person’s mental health.
Sign up for Chip Chick’s newsletter and get stories like this delivered to your inbox.
Furthermore, new data released by OpenAI, the company that runs the ChatGPT chatbot, shows that a small percentage of users exhibit signs of mental health emergencies related to mania or psychosis each week.
At least 13% of Americans between the ages of 12 and 21 are turning to generative AI for advice on mental health. In the 18 to 21 age group, which are the peak years for the onset of psychosis, the percentage is even higher at 22%.
AI chatbots reinforce delusions because they are essentially sycophants. They are programmed to be people pleasers and tell users what they want to hear.
“I’m especially concerned about lonely young people who are isolated and thinking that their only friend is this chatbot, when they don’t have a good understanding of how it’s behaving or why its programming might lead it to react in certain ways,” said Stephan Taylor, a psychiatrist and professor at the University of Michigan.
If someone chooses to use AI tools or chatbots regarding their mental health, it’s important that they also talk to a trusted human about their feelings, whether that person is a therapist, friend, relative, or teacher.
The study mentioned above can be found on the preprint server PsyArXiv.
If you or someone you know is struggling, please reach out to a human. You can call or text 988 in the US and Canada to reach the Suicide & Crisis Lifeline, or visit your local emergency room for immediate help.
By Emily Chan
Emily Chan is a writer who covers lifestyle and news content. She graduated from Michigan State University with a degree in... More about Emily Chan
