OpenAI (the company behind ChatGPT) is implementing a series of improvements to encourage the healthy use of this technology.
Recently announced changes include updating chatbot feedback behavior, integrating rest reminders, and guiding users to evidence-based information sources in case of emotional crisis, according to indianexpress.
Adjust feedback in sensitive situations
OpenAI said it is collaborating with healthcare professionals, clinicians, researchers and psychological counselors to help ChatGPT better detect and respond to signs of mental illness from users.
The goal is to make AI chatbots not only safer, but also more useful in sensitive moments.
For example, with questions like Should I break up with my lover?, ChatGPT will not give a direct answer. Instead, the chatbot will ask suggestive questions, helping users consider their pros and cons and make their own decisions.
This is to avoid falling into a situation where AI gives decisive advice that is not suitable for the personal circumstances of users.
Integrating reminders for rest in long chat sessions
A new feature that has just been enabled by OpenAI is a gentle reminder for those who have been chatting continuously for a long time.
The window that bounces up in a soothing tone will display a message like: "You've been chatting for a while, now you're comfortable resting?", with two options: "Continuing the conversation" and "This information is useful".
Similar to social media platforms like YouTube and Instagram, this reminder helps reduce technology abuse, especially for those looking for comfort from AI rather than from human resources.
The fact that users consider ChatGPT as a virtual therapist has raised concerns in the psychological world. Many experts warn against AI, which is designed to please users but can unintentionally reinforce negative thinking, even leading to serious consequences if not controlled.
OpenAI admits that some previous updates have made chatbots too profitable, leading to slanderous or unnecessary responses. The company said it has overcome and is improving the criteria for measuring practical usefulness, instead of relying only on the temporary " likes" feeling from users.
Towards a safer future with GPT-5
OpenAI is expected to launch the GPT-5 language model this week, promising to continue improving the accuracy, humanity and empathy of AI.
With more than 700 million active weekly users, promoting responsible and healthy ChatGPT usage is becoming a top priority for OpenAI.
