China put robots in a cage, and it was not for entertainment

In this monologue, Dan Novaes describes a new wave of humanoid robots battling in cage arenas, likening the spectacle to a modern version of BattleBots. He explains that every movement these robots perform—kicks, flips, dances—is learned from large datasets of human motion, a challenging problem in physics and AI. Novaes notes that China is deliberately showcasing these robot fights in public arenas as a tactical move to demonstrate its advanced robotics capabilities and signal what’s coming next. He argues that physical AI, trained on human data, will soon extend beyond combat to household chores and blue‑collar work, and that individuals can now be compensated for contributing their motion data. The video serves as a warning and a preview of how physical AI could reshape entertainment, labor, and data monetization.











