Early Methodologies for Researching Emotional Use of ChatGPT and Emotional Well-being
Key point
OpenAI and MIT Media Lab conducted research analyzing the impact of emotional use of ChatGPT on users' well-being.
Details
People use ChatGPT not just for simple information retrieval but also as a tool for personal interaction. Accordingly, research was conducted to understand the impact of AI interactions on users' social and emotional well-being.
OpenAI and MIT Media Lab conducted two types of research to understand the impact of 'affective use' on users.
- Study 1 (OpenAI): To protect user privacy, approximately 40 million ChatGPT interactions were automatically analyzed without human involvement. This was combined with user surveys to analyze the correlation between actual usage patterns and emotional use.
- Study 2 (MIT Media Lab): A 4-week randomized controlled trial (RCT) was conducted with approximately 1,000 participants. The study investigated the causal relationship between specific features such as the model's personality or modality and loneliness, social interaction, and emotional dependence on AI.
The research findings showed that emotional engagement through ChatGPT was a very rare occurrence in real-world service environments. Interactions containing emotional expression were concentrated among a small group of heavy users who frequently used Advanced Voice Mode, rather than the majority of users. This group showed a relatively higher tendency to regard ChatGPT as a friend.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.