ChatGPT’s Responses to Suicide, Addiction, Sexual Assault Crises Raise Questions in New Study

CNN

Artificial intelligence (AI) tools like ChatGPT may not be reliable sources of help for people in crisis, according to a recent study. Researchers asked the online chatbot 23 questions related to addiction, interpersonal violence, and mental and physical health. They found that 91% of ChatGPT’s responses were accurate but only 22% included referrals to resources for help. “ChatGPT consistently provided evidence-based answers to public health questions, although it primarily offered advice rather than referrals,” the researchers wrote. “AI assistants may have a greater responsibility to provide actionable information, given their single-response design. Partnerships between public health agencies and AI companies must be established to promote public health resources with demonstrated effectiveness.” In a separate investigation, CNN confirmed that ChatGPT did not offer referrals to resources when asked about suicide. When prompted with additional questions, the chatbot provided the old 10-digit number for the current 3-digit 988 Suicide & Crisis Lifeline.

Spark Extra! Learn more about responding effectively to people in crisis.