Next Article
OpenAI merges teams to make ChatGPT smarter, friendlier
OpenAI just merged its Model Behavior team (the folks who work on personality and bias issues) with the Post Training group, with the combined team now reporting to Post Training lead Max Schwarzer.
This move is meant to blend personality tweaks with model improvements, so future AIs can be both smarter and friendlier.
Now, the combined team reports to Max Schwarzer.
GPT-5's chilly replies prompted user feedback
This change comes after users said ChatGPT (especially GPT-5) was starting to sound a bit cold. OpenAI took that seriously and updated the models for warmer, more natural replies.
Plus, Joanne Jang—who started the Model Behavior team—is launching OAI Labs, a new project exploring fresh ways for people and AI to work together beyond just chatting.