MLOps.community The Missing Data Stack for Physical AI
21 snips
Jul 1, 2025 Nikolaus West, CEO of Rerun and a Stanford computer vision alum, dives into the intriguing world of physical AI and robotics. He discusses how traditional software struggles in dynamic environments and emphasizes the need for better visualization and adaptability tools. West explores how physical AI could revolutionize industries, contrasting it with the stagnation of IoT. He also highlights the complexities of training AI, the challenges of data management in robotics, and the importance of balancing sensor types and modeling approaches in creating effective solutions.
AI Snips
Chapters
Transcript
Episode notes
Expansive Scope of Physical AI
- Physical AI broadly applies AI to interact with the real world in various forms like robotics and spatial computing.
- It encompasses many applications beyond traditional robotics, shaping the physical economy through intelligent interaction.
IoT vs Physical AI Impact
- IoT hype failed to transform real-world applications significantly over decades.
- Unlike IoT, physical AI offers value by performing actual work and understanding in the physical world.
AI and Scale Enable Physical AI
- AI's capability to handle real-world complexity and ambiguity enables physical AI breakthroughs.
- Scale in hardware and data collection creates a flywheel effect, accelerating advances and lowering costs for physical AI products.