Archive — US — techcrunch
OpenAI reorganizes research team behind ChatGPT’s personality
OpenAI is reorganizing its Model Behavior team, integrating its functions into the larger Post Training team. This move is intended to enhance the interaction and personality of AI systems following issues with user responses, particularly regarding sycophancy and the recent case linked to a tragic event. Former leader Joanne Jang is launching OAI Labs to explore new AI collaboration interfaces.
Key points
- OpenAI's Model Behavior team is being merged into the Post Training team to improve AI models' interaction with users.
- The Model Behavior team, consisting of around 14 researchers, aims to shape AI personalities and reduce sycophancy.
- Joanne Jang, former leader of the Model Behavior team, is starting OAI Labs at OpenAI to develop new AI collaboration interfaces.
- User feedback on GPT-5 prompted OpenAI to adjust AI personalities to be friendlier while minimizing sycophancy.
- A lawsuit against OpenAI highlights the potential risks of AI interactions, emphasizing the need for responsible AI response management.
Why it matters
- Increased scrutiny on AI models following a tragic case linked to their responses highlights the ongoing debate about AI responsibility.
- Trends in AI focus on enhancing user engagement while balancing ethical considerations, a key challenge for AI development toward 2025.
- Developments in AI user interfaces are a growing trend, with companies like OpenAI exploring innovative ways for better collaboration.
- The structure changes in AI companies reflect broader shifts within tech firms to improve AI models amidst rising global concerns about AI ethics and effectiveness.
OpenAI's restructuring of its Model Behavior team is a strategic response to user feedback and ethical concerns, aimed at fostering a more balanced AI interaction experience. The formation of OAI Labs by Joanne Jang signifies a forward-looking approach to AI collaboration and innovation.