Rohan Paul(@rohanpaul_ai):🧵 4. Human-centric Paradigm，AWE 3.5 was trained by over 1 million hours of human-centric data. TARS proposes a human-centric data-collection paradigm as an essential path to scaling data for physical AI. The human-centric paradigm is designed to collect data from human in real world tasks.The data acquisition system covers diverse scenarios , supporting the development of large-scale physical AI capabilities.

Rohan Paul

@rohanpaul_ai

Compiling in real-time, the race towards AGI. 🗞️ Get my daily AI analysis newsletter to your email 👉

Joined June 2014

6.8K Following 158K Followers

Rohan Paul@rohanpaul_ai

2026.08.21 18:09

🧵 4. Human-centric Paradigm，AWE 3.5 was trained by over 1 million hours of human-centric data. TARS proposes a human-centric data-collection paradigm as an essential path to scaling data for physical AI. The human-centric paradigm is designed to collect data from human in real world tasks.The data acquisition system covers diverse scenarios , supporting the development of large-scale physical AI capabilities.