Huang Biwei

Huang Biwei is the Aether AI founder interviewed in 哪条路线，才能通往「世界模型」的终局？｜对话黄碧薇：Aether AI 创始人. The source presents her as a long-time Causal AI researcher with experience at institutions including the Max Planck Institute, CMU, and UCSD, and as an entrepreneur trying to turn causal research into a robot-brain product.

Thesis

Huang argues that World Models only become practically useful for physical tasks when they learn latent causal variables, causal structure, and action-conditioned transition dynamics. Her strategic choice is to focus on Embodied AI because language and code tasks are already relatively mature under the LLM paradigm, while robotics still needs a deeper model of physical regularities.

Connections

• Aether AI — company she founded around causal world models for embodied intelligence.
• Causal World Models — core technical route she advocates.
• Causal AI — research field that shaped her work.
• Vision Language Action Models and World Action Models — robot-modeling approaches she treats as useful but incomplete.
• AI For Science — broader set of physical and scientific domains where she expects causal modeling to matter.