General-purpose LLM beats specialized robot models on arm control, signaling robotics' bitter lesson
zakkohane · x · 2026-11-02
Zak Kohane argues the bitter lesson is arriving in robotics: a general-purpose LLM (ChatGPT Astra) given a robotic arm and camera outperforms state-of-the-art specialized robot models trained on thousands of hours of video. Key enabler is "geometric prompting" — supplying the arm's specs unlocks fast learning, with big implications for RL transfer in domains like medicine. His headline takeaway: orchestrated agents underperform an unconstrained agentic policy.
More from Embodied
- Bleu Robotics raises €5M pre-seed to deploy humanoids in factories with just 10 demos per task — lukas_m_ziegler · 2026-11-03
- Machenta exits stealth with $40M seed, largest ever for a UK robotics startup — lukas_m_ziegler · 2026-11-03
- World model that predicts the future directly in 3D, fully open-sourced — YunzhuLiYZ · 2026-11-03
- Dream-1 zero-shot robot test: 100 public episodes show real skills but poor instruction following — paigeinsf · 2026-11-03
- ANYbotics ships ANYmal X, the first legged robot certified for explosive atmospheres — lukas_m_ziegler · 2026-11-03
- Tesla starts production-line Optimus build at Fremont, external deliveries eyed for late 2027 — lukas_m_ziegler · 2026-11-03