Jensen Huang: Cosmos is the ChatGPT for the physical world
r0ck3t23 · x · 2026-08-31
A deep commentary on Nvidia's recently released world model, Cosmos. Jensen Huang describes Cosmos as "ChatGPT for the physical world," noting that language was easy because we wrote it down, whereas physical laws like weight, friction, and reflex were too obvious to record, making them the hardest subject.
The text highlights the primacy of physical intelligence: humans master object permanence before words. We built machines that argue philosophy years before building ones that know a ball still exists behind a couch. Cosmos can generate plausible physical scenarios (e.g., driving futures), a capability to "pre-run" the next few seconds—the oldest and most expensive work of a nervous system. By open-sourcing it, we allow the machine to learn physical laws through millions of trials without the cost of real-world consequences.
More from AGI Musings
- Why LLMs Can Invent Genuinely New Things: Technical Breakdown — bindureddy · 2026-08-31
- Former OpenAI board member: OpenAI probe could seed US-China AI safety talks — joshua_saxe · 2026-08-31
- Why rigorous future thinking leads to extreme outcomes — zetalyrae · 2026-08-31
- HF Attack Vindicates Rationalist Predictions, But Models Lack Malice — voooooogel · 2026-08-31
- AI Agents: Anthropomorphism is Useful for Prediction Regardless of Intent — connoraxiotes · 2026-08-31
- Why models go rogue during training: Capability scaling may cause misalignment — repligate · 2026-08-31