Physical AI
Also written: Physical intelligence
AI that can control robots and interact with the real world.
Physical AI refers to artificial intelligence systems designed to operate in the physical world. This means the AI can control robots or other machines to perform actions like moving objects, navigating spaces, or manipulating tools.
Think of a robot arm in a factory assembly line that uses AI to pick up and place parts precisely, or a humanoid robot that can walk around a room and pick up a dropped item. This type of AI needs to understand physical properties like gravity, friction, and object stability.
Physical AI involves AI systems that are embodied or control robotic systems to interact with the physical environment. This requires AI models to process sensory data (vision, touch, proprioception) and generate motor commands forctuation. Key challenges include real-time control, dealing with uncertainty and variability in the physical world, and learning complex manipulation and locomotion skills. Current research focuses on integrating large language models and other foundation models with robotic control systems to enable more general-purpose physical interaction.
AI is moving beyond software to control physical systems like robots and autonomous vehicles. Companies are investing in AI that can interact with and act in the real world, leading to smarter automation and more capable robots. This shift is driven by advancements in robotics and the need for AI to perform complex physical tasks.
- From Anthropic to Waymo: Kleiner Perkins Bets on AI’s Next ActAnthropic · Bloomberg Technology ·
- The next step in factory automation: Machines that adapt with AIMicrosoft · Microsoft Source ·
- Offloaded inference for real-world physical AI roboticsMicrosoft · Microsoft Research Blog ·
- Robots are waiting for a ChatGPT moment: Nvidia’s Les Karpas explains why at TechCrunch Disrupt 2026OpenAI · TechCrunch AI ·
- Tasks that require whole-body control with Gemini Robotics 2Google · Google DeepMind on YouTube ·
- Intelligent whole-body control with Gemini Robotics 2Google · Google DeepMind on YouTube ·
- Let's get physicalGoogle · Google DeepMind: The Podcast ·
- Anthropic Claude
- Anthropic is seeing venture capital investment in physical AI applications as the next major wave.
- Google Gemini
- Google Gemini is advancing physical AI with its robotics efforts, enabling complex whole-body control for robots.
- Microsoft Copilot
- Microsoft Copilot is involved in developing AI for factory automation and enabling robots to perform complex tasks through offloaded inference.
- OpenAI ChatGPT
- OpenAI's work is seen as a benchmark, with discussions around robots needing a similar 'ChatGPT moment' to advance.
- Artificial General Intelligence (AGI)
- AGI refers to AI with human-level cogniti
- Robotics
- Robotics is the field of designing, building, and operating robots, wh
The term 'physical AI' emerged as AI capabilities expanded beyond purely digital domains into the physical realm, particularly with the rise of advanced robotics and embodied AI research. It signifies AI's transition from software to hardware interaction.
While direct interaction with physical AI robots is limited, you can explore related concepts. Try using a robot simulator like CoppeliaSim or Gazebo to experiment with robot control. Alternatively, interact with AI-powered tools that have physical world implications, such as AI image generators that can be used to design physical objects.
The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news.
- ChecklistDesign considerations for physical AI in manufacturingKey factors to design into AI systems operating in the physical world. · 20 min
- PromptRole-play a student struggling with math for AI tutor simulationSimulate a student's math struggles and uncertainty for AI tutor evaluation. · 5 min
- TemplateReusable workflow templates for healthcare teamsTeams can create and share templates for common workflows like documentation or prior authorizations. · 15 min
- TemplateTemplate for building an AI tool tailored to your workA structured approach to defining and building a custom AI tool for specific business needs. · 15 min
- TemplateEvals CSV prompt list for agent skillsA CSV file template for listing prompts used in evaluating agent skills. · 20 min
Questions people ask
- What's the difference between physical AI and regular AI?
- Regular AI often operates purely in the digital world, like chatbots or recommendation systems. Physical AI is specifically designed to interact with and act within the real, physical world, usually by controlling robots.
- Do I need a robot to use physical AI?
- Yes, physical AI typically requires a physical embodiment, such as a robot, to interact with the real world. The AI provides the 'brain' that controls the robot's actions.
First seen on Subvolts Feb 8, 2022. Updated Oct 16, 2026. Written by an AI model from the items above and reviewed against them weekly; report an error.