Are brain waves the next unlock for physical AI?

AI Summary

Encord, an AI data tooling startup, is experimenting with brain wave measurements to improve physical AI training data. Working with German neuroscience firm Zander, Encord equips human robot trainers with EEG headsets to capture mental states like intent, error, and surprise alongside standard motion data. The company argues that the primary bottleneck for humanoid and warehouse robotics is not model architecture but the severe scarcity of real-world physical training data. Unlike LLMs built on vast amounts of freely available internet text, physical AI requires expensively manufactured, densely annotated data sets. Encord is pursuing multiple data collection methods, including egocentric video, leader-follower robotic arm demonstrations, and forearm muscle-signal sensors, while serving numerous unnamed robotics clients. Its cross-industry visibility gives it a unique ability to identify which data techniques are gaining traction before individual companies can.