Encord is pioneering the use of brain wave data to enhance the training of physical AI systems. The company operates in a warehouse in San Leandro, California, where it collects egocentric data through specialized sensors and headsets. This data is crucial for training AI models that interact with the physical world, such as robots performing tasks like manipulating objects or navigating complex environments. Encord’s approach involves capturing not just visual data but also neural signals, which could provide deeper insights into human intent and movement.

The company’s pilot program involves Andrew Ceja, who wears a headset equipped with cameras and sensors to collect training data. This method is part of a broader trend in the AI industry, where companies are increasingly relying on egocentric data—data collected from the first-person perspective of workers or users. Encord’s data collection process includes not only video but also detailed annotations that capture the precise movements and intentions of individuals interacting with objects. This approach is seen as a more efficient and accurate way to train AI models compared to traditional methods.

By 2026, Encord expects its data collection and annotation methods to scale significantly. The company estimates that the dense annotation of data, which captures detailed information about hand movements and object manipulation, is worth 100 times as much as less precise data. This is a significant advantage, as the cost of producing this high-quality data is only 20 times higher than that of less detailed data. This ratio makes the investment in high-quality annotation a compelling proposition for companies looking to train AI models with greater accuracy and efficiency.

The use of brain wave data and advanced annotation techniques could have wide-ranging consequences for the AI industry. Companies may face increased costs associated with data collection and annotation, but the potential gains in model accuracy and performance could justify these expenses. Additionally, the reliance on specialized data could lead to vendor lock-in, as companies may become dependent on specific tools or platforms for data collection. Governance and ethical considerations will also become more important as the use of brain wave data raises questions about privacy, consent, and the potential misuse of neural signals.

Encord’s approach is still in its early stages, but it represents a significant shift in how physical AI is trained. The company’s focus on high-quality, annotated data may set a new standard for the industry. As the technology continues to develop, it will be important to monitor how these advancements impact the broader AI ecosystem, including the cost of data, the efficiency of training models, and the potential for new applications in robotics and automation.