Definition
Physical AI is an umbrella term for AI systems that take geometry, material, machines or physical processes as input or output, rather than only language or images. Robotics, autonomous vehicles and systems that understand a physical part all fall under it.
What it means for the calculation
A world model for precision manufacturing is a form of physical AI: it works with a part's geometry, a machine's capability and the process that produces a physical object, not with text on its own. That sets the task apart from a language model that only generates text.
Where this comes up
In any system that treats geometry, material or machine behaviour as a first-class input, rather than as a caption to text.
Common mistakes
Confusing physical AI with any AI system that happens to talk about factories; the distinction is in what the system actually takes as input, not in the subject of the text around it.
A world model for precision manufacturing is a form of physical AI: it works with a part's geometry, a machine's capability and the process that produces a physical object, not with text on its own. That sets the task apart from a language model that only generates text.
Read the researchFrequently asked questions
Is physical AI the same as robotics?
Robotics is one application of it; physical AI is the broader category that also covers understanding geometry and process.
Is Blake physical AI?
Blake works with a part's geometry and machine data, so it falls in this category; it is not robotics and does not control a machine.
What is the difference from a language model?
A language model works primarily with text; physical AI has geometry, material or machine behaviour at the core of its input or output.