OpenAI is changing how ChatGPT interacts with users, especially for sensitive topics like breakups. Instead of giving direct answers, the AI will now help users think through their decisions by prompting them with questions to consider the pros and cons.
The goal is to avoid giving potentially harmful advice about high-stakes personal issues. OpenAI acknowledged that earlier versions of ChatGPT sometimes failed to recognize signs of emotional distress, leading to concerning interactions. This raised red flags about how AI might affect mental health.
Recent research from NHS doctors in the UK highlights these issues. The study found that AI chatbots can sometimes mirror and amplify delusional thoughts in users vulnerable to psychosis. It emphasizes a need for caution, as AI’s focus on engagement may blur the lines between reality and fiction.
As part of its new approach, OpenAI plans to implement reminders for users who spend too much time chatting. This is similar to screen-time alerts seen in social media platforms. Together with feedback from mental health experts, OpenAI aims to create a system that is safe and supportive.
The company has gathered insights from over 90 medical professionals to refine its chatbot interactions. Their goal is simple: if a loved one were to seek advice from ChatGPT, they want to feel confident that it would provide appropriate support.
These updates come as speculation grows about a more advanced version of the AI, GPT-5. OpenAI is focused on ensuring that any enhancements prioritize user wellbeing.
For more on the challenges and implications of AI in mental health, you can read more here.
