OpenAI is restructuring its influential Model Behavior team, consisting of about 14 researchers, integrating them into the larger Post Training group led by Max Schwarzer. This shift aims to enhance the interaction quality of OpenAI’s models, including addressing concerns about AI sycophancy and biases. Joanna Jang, the previous head of the Model Behavior team, is launching OAI Labs to explore innovative interfaces for AI collaboration. Recent public scrutiny arose following a lawsuit related to ChatGPT and its handling of sensitive topics, highlighting the importance of balancing user engagement with responsible AI responses. OpenAI’s ongoing work—including updates to models like GPT-5—shows a commitment to making AI “warmer” while minimizing biases. Jang’s enthusiasm for developing beyond traditional chat paradigms reflects a broader vision for AI as a collaborative tool rather than merely a conversational agent. The evolution of these teams signals a pivotal moment in AI development and user experience.
Source link
Share
Read more