Definition
Physical AI describes AI systems that perceive, understand and act in the physical world through embodiment. The term was popularised by NVIDIA for robots, autonomous machines and humanoids. What defines the category is that the system operates through actuators, motors or wheels, not that it discusses physical objects.
Why Blake is not physical AI
A world model for precision manufacturing is not physical AI. It reasons over a part's geometry, requirements and factory context to produce a reviewable calculation, but it does not move a machine. The correct category for Blake is physics-informed AI: algorithms and mathematical models that build physical laws into how they reason.
Where physical AI applies
In any system that acts on the world through a body: robotics, autonomous vehicles, humanoids.
Common mistakes
Confusing physical AI with physics AI. Physics AI, short for physics-informed AI, is about how a model reasons, with physical laws built into the algorithm. Physical AI is about what a system does, acting through a body. The names sound alike; the categories are different.
A world model for precision manufacturing is not physical AI. It reasons over a part's geometry, requirements and factory context to produce a reviewable calculation, but it does not move a machine. The correct category for Blake is <a href="/glossary/physics-informed-machine-learning">physics-informed AI</a>: algorithms and mathematical models that build physical laws into how they reason.
Read the researchFrequently asked questions
Is Blake physical AI?
No. Blake reasons about a part and produces a calculation for a human to review; it does not move a machine. The accurate framing for Blake is physics-informed AI, sometimes called physics AI in shorthand.
What is the difference between physical AI and physics AI?
Physical AI acts through a body: robots, autonomous vehicles. Physics AI, short for physics-informed AI, is algorithms and mathematical models that embed physical laws in how they reason.