OpenAI has released new estimates of users exhibiting possible signs of mental health emergencies, including mania, psychosis, or suicidal thoughts on its platform.
The company states that 0.07% of active users (around 560,000 people) showed these symptoms, which it considers “extremely rare.” However, critics argue that even a small percentage can add up to hundreds of thousands of people, considering ChatGPT has 800 million weekly active users.
To address this, OpenAI has built a network of experts around the world, including over 170 psychiatrists and psychologists from 60 countries. They’ve developed responses to encourage users to seek help in real life.
Mental health professionals have raised concerns about AI’s limitations in supporting mental health. Dr. Jason Nagata says that even though 0.07% may seem small, it can impact many people at a population level. He emphasizes the need for awareness of AI’s capabilities and limitations.
OpenAI has made updates to its chatbot to respond safely and empathetically to signs of delusion or mania and indirect signals of potential self-harm or suicide risk. The company also reroutes sensitive conversations to safer models by opening in a new window.
These changes come as OpenAI faces mounting legal scrutiny over the way ChatGPT interacts with users, including lawsuits alleging that the platform contributed to user deaths.
Source: https://www.bbc.com/news/articles/c5yd90g0q43o