OpenAI recently made a big move by merging the model behavior team with the post-training team. It seems they want to closely integrate AI's personality with technical development. Quite interesting.

The model behavior team originally had around 14 researchers responsible for shaping how OpenAI's AI interacts with people. Now, they are being merged into the post-training team. The new boss is Max Schwarzer.

There has been quite a change. Joanne Jang is starting a new project called OAI Labs. She said she wants to "invent new interface prototypes" for collaboration between humans and AI. It sounds pretty impressive.

Mak Chen emphasized in an internal memorandum that now is a good time. To closely integrate model behavior with core development. It seems that OpenAI believes that the personality of AI is very important.

Recently, OpenAI has faced quite a bit of criticism. After GPT-5 was released, some people said it became aloof. OpenAI made some adjustments, aiming to make it both friendly and not obsequious. It's not easy.

The modeling behavior team has participated in several major projects. GPT-4 and GPT-5 both have their influence. Jang also participated in Dall-E 2 before. Quite impressive.

Jang wants to explore new paradigms and tools in the new project. It's still unclear how exactly to do it. But she emphasizes the need to go beyond traditional models.

This restructuring seems to indicate that OpenAI is adjusting its strategy. It is placing more emphasis on the behavior and personality of AI. Hopefully, it can create a better human-machine interaction experience.