What Is Physical AI? How Is It Different from Generative AI? 10 Questions for Jun Yamaguchi at the Kawasaki Physical AI Center

A. In the physical AI space, our center brings three things together: Silicon Valley's AI development capability and speed, Kawasaki's product lineup — starting with robotics — and our accumulated know-how about markets, customers, and real-world operations. By combining them, we aim to open up entirely new business areas.

A. Until recently, AI has mostly lived inside our screens — in the virtual space of smartphones and PCs. Generative AI, like ChatGPT, has made it possible to have smooth conversations and to generate text and images.

Physical AI is the natural next step: AI that reaches out from the screen into the real, physical world. Think of self-driving cars, or robots — physical AI can recognize what's happening around it, make decisions, and move things in the real world. As it spreads, we can expect the machines around us to move more flexibly, making our daily lives more convenient.

A. San Jose sits at the heart of Silicon Valley — a place where investment and talent gather from all over the world. It's often called "the place where the world's next decade is decided," and right now, the focus of that decade is physical AI.

Kawasaki has been operating in San Jose since 2001, building it into our U.S. robotics business base. That means we already have easy access to the robotics hardware that physical AI needs to become real. To turn this advantage into business growth, we opened the Physical AI Center here.

A. Since starting our robotics business in 1968, Kawasaki has stayed close to our customers, allowing us to accumulate know-how and data from real worksites. That data is exactly what physical AI needs to learn from. It's something that companies without hardware, or those that provide only general-purpose hardware, simply cannot offer.

We also cover a wide range of robots — from industrial robots to social robots that work alongside people. In manufacturing, for example, we've analyzed tasks like welding and painting, and used that on-the-ground knowledge to build robots. More recently, we've turned our attention to highly skilled professionals — like doctors — analyzing their movements to figure out which parts of the work can be shared with robots. Being able to work from both angles — the "task" side and the "human skill" side — is another strength Kawasaki brings to physical AI.

A. Rather than adding AI product by product, we're thinking in terms of solutions: multiple products working together as one. In hospitals, for instance, we position hinotori™, Nyokkey, and FORRO together as a "one-stop solution" that supports both patients and healthcare staff. AI helps these robots coordinate with each other and with hospital systems, and respond intelligently to what patients and staff actually need.

Kaleido, meanwhile, is being designed as a hybrid — combining conventional motion-control software with AI. Movements that need reliability are handled by traditional control, while adapting to change and making decisions is where AI comes in.

(*) hinotori is a trademark of Medicaroid. Medicaroid is the legal manufacturer of hinotori™ and a joint venture between Kawasaki and Sysmex.

A. CORLEO is an RMV (Robotic Multi-legged Vehicle) — an entirely new category of vehicle that no one in the world has experienced before. So, the biggest challenge is determining how it should feel to ride and how it should handle. Our answer is a "virtual off-road course" built with AI, where we can simulate operation and control systems thousands, even tens of thousands, of times. That lets us push hardware development forward safely and efficiently.

We're also planning to use AI to read and predict road conditions during actual runs, and to help close what's called the "Sim2Real gap" — the difference between how CORLEO behaves in simulation and how it moves in the real world.

A. Physical AI should help soften the impact of a shrinking working population caused by aging societies. We expect it to become a natural part of manufacturing, service industries, and healthcare — quietly supporting people through AI-powered equipment and robots.

Its use will expand into homes, too. In healthcare, physical AI could help with health monitoring and daily support — potentially delaying the need for hospitalization, or speeding up discharge, and easing the burden on healthcare institutions. All of this should help older adults live more safely, more comfortably, and more independently.

Physical AI will also be valuable in times of natural disaster and other emergencies.

A. In San Jose, working with partner companies and taking advantage of Silicon Valley's speed, we cycle through rapid prototyping — building, testing, iterating — to validate technologies and services on the path to making physical AI real.

Those results move to Europe, where the focus is on real-world implementation in medical and healthcare sectors, and gathering feedback from the field. That feedback comes back to San Jose, feeding into the next round of prototyping. It's a continuous evolution loop.

Japan, meanwhile, serves as our primary hub — driving the broader expansion of robotics into all kinds of fields, and creating new markets even beyond robotics.

A. Our ambition is to become the kind of hub that can ride the next big wave — the "world's next decade" that starts here in Silicon Valley.

We want to catch the next trends early, act as the spark for Kawasaki to expand into new markets and create markets that don't yet exist, and earn a reputation as "the place to ask about what's coming next."

A. Right now, generative AI is quietly becoming part of everyday life — from writing to image creation. Going forward, physical AI will take another step: it will understand what's happening around it on its own, decide what to do, and react — without waiting for our instructions.

As a manufacturer, we want to expand the possibilities of physical AI together with society. That's exactly why we want to listen carefully to how you feel about it, and what you hope for — so that we can bring physical AI into daily life carefully, thoughtfully, one step at a time.