OpenAI is set to enhance ChatGPT with features aimed at curbing unhealthy user behavior. Starting Monday, the app will issue “gentle reminders” encouraging users to take breaks from lengthy interactions. OpenAI acknowledged the limitations of its GPT-4o model in identifying emotional dependency and delusional behavior, stating that it is developing tools to detect mental distress effectively. This initiative responds to concerns over instances where ChatGPT’s responses led users to dangerous situations, including suicidal thoughts. OpenAI is collaborating with over 90 physicians and forming an advisory board of mental health experts to refine conversational guidelines. CEO Sam Altman highlighted privacy issues regarding sensitive discussions with AI, advocating for confidentiality akin to therapist-client relationships. The company’s focus is on improving user experience by encouraging shorter engagements while ensuring ChatGPT remains a supportive tool. They aim to redefine success metrics based on user satisfaction rather than interaction time.
Source link
Share
Read more