"Artificial intelligence today is overwhelmingly associated with virtual tools - LLMs, chat interfaces, and digital agents that live entirely on screens. But as Avinash Balachandran, Vice President of the Human Interactive Driving (HID) Division at Toyota Research Institute (TRI), emphasized during his keynote at ODSC AI West 2025, the next frontier of AI won't be found online. It will be found in the physical world."
"Most people today meet AI through chatbots, image generators, or recommendation systems. These tools help with planning, creativity, and information retrieval, but they don't help us chop vegetables, fold laundry, or keep a car under control in a snowstorm. Balachandran noted that this distinction - between digital and physical assistance - is precisely what embodied intelligence aims to bridge. These systems aren't trained only on internet-scale data; they also learn from physical interaction, just as humans do."
Embodied intelligence integrates AI into physical machines—cars, robots, humanoids—that interact directly with people and environments. These systems learn from physical interaction in addition to internet-scale data, enabling assistance with tasks like chopping vegetables, folding laundry, and maintaining vehicle control in adverse conditions. The aim is to amplify human intelligence, capability, and confidence rather than replace people. Research and development prioritize autonomy and robotics within the context of building real-world physical machines. This shift creates new opportunities, introduces new risks, and alters the human-machine relationship.
Read at Medium
Unable to calculate read time
Collection
[
|
...
]