Robots That Predict the Future: Dyna-2 & Physical AI — Jason Ma, Dyna Robotics
Latent Space published this 1-hour-21-minute-5-second interview on October 13, 2026; host swyx talks with Dyna Robotics co-founder and Chief Scientist Jason Ma about the Dyna-2 world-action model: training on over a million hours of human video, synthetic data, the wheeled semi-humanoid hardware cho
Key points Tap a time to jump
Introduction and Dyna-2's world-action model
Diffusion transformers, flow matching, and generalist robots
The robotics data bottleneck
Synthetic data and self-improving world models
Full-stack robotics: hardware and supply chains
Why Dyna built a wheeled semi-humanoid
Autonomous workflows and hardware limitations
Real-world deployments at Best Western and Din Tai Fung
The bitter lesson of robotics
GPT-6, coding agents, and AI-native robotics
Comments
More to watch
All →- 1:34:05All-In Podcast E293: CoreWeave Crash, AI Trade Cracks, Anthropic IPO Red Flags, Bill Gates Doomer
- 39:23No Priors with OpenAI President Greg Brockman: Cybersecurity, Drug Discovery, and the Math Results
- 1:49:56Why a 4-Billion-Point Kaggle Dataset Still Couldn't Beat Chance — Ian Quigley, Leash Bio (Latent Space)
- 28:04How to kill your SaaS: what we learned replacing a $43K/yr tool with AI agents
Latest news
All →- Oct 18Anthropic to Commit 5% of Shares to New Public Benefit Fund
- Oct 19Stably AI's Orca Tops GitHub Daily Trending: One Desktop App to Orchestrate 30+ Coding Agents
- Oct 19Jailbroken PS5 Becomes a Local AI Server: Qwen3.8-27B Hits ~20 tok/s, Small Models Top 100
- Oct 19Anthropic Internal Study: Junior Share of Technical Staff Falls to 14%, Meeting Time Up 20%
- Oct 18Claude Cracks 25-Year Cryptography Mystery in Half an Hour: NIST Curve Seeds Came From a 1997 Clock
- Oct 18OpenCode Hit by "Distillation Attack": API Keys of About 1.4% of Users Compromised
Today
Oct 19 Monday- Brief
- GitHubGitHub Weekly Vol.1: 8 Projects Worth Your Time
Tested or read — weeklyAll →
- Quiz
- Call
™