LLMs zero-shot robot arm control is unsurprising; dexterity still favors specialized models
adam_karvonen · x · 2026-11-01
Adam Karvonen comments on the OpenAI robotics result: it's unsurprising that a strong LLM can do basic pick-and-place tasks, since humans can do the same with a joystick and zero practice, and such tasks are easy to train in simulation. Dexterity-heavy tasks like screwing in a lightbulb or operating a humanoid hand remain where specialized robotics models are far better.
More from Embodied
- Naval predicts 4-passenger drones will eventually end commercial air travel, with a long interregnum ahead — naval · 2026-11-01
- Meta Quest game scans your real room and turns it into a haunted house — Scobleizer · 2026-11-01
- RoboAuto: Robot Program Search With Multi-Level Physical Understanding — DJiafei · 2026-11-01
- Developer dresses robot 'Bob' as a ghost to hand out Halloween candy — chrismatthieu · 2026-11-01
- GRASP authors compare with SuperMimic: extracting more optimization signal from sim experience — hankyang94 · 2026-11-01
- Harvard's GRASP refines human demos in physics sim for dexterous hands and Unitree G1 — hankyang94 · 2026-11-01