AI Tech / news
Brain Wave Readings Enter Physical AI Training Data Arsenal
Encord and Zander Labs are testing brain wave headsets to capture mental states like error and intent, aiming to enrich training data for humanoid and warehouse robots.
In a San Leandro, California warehouse, Encord's pilot Andrew Ceja plays Jenga while wearing a headset that measures his brain waves. The setup is part of a broader push to generate scarce physical-world training data.
Zander Labs, a German neuroscience startup, built the headset. Their collaboration reflects a belief that the next bottleneck in robotics is data scarcity, not model architecture.
Brain activity readings can reveal when a human detects errors or feels surprise, offering cues for AI models to optimize where to apply computational effort. Lucas Gehrke, a Zander neuroscientist, noted that the level of brain activity during tasks helps determine when high-effort models are needed.
Encord is one of a small but growing number of startups betting that the next real constraint on humanoid and warehouse robotics will be the sheer scarcity of real-world physical training data, and is building a business around manufacturing the data that does not yet exist.