Robotics veteran Jitendra Malik challenges LLM robotics hype with a dexterity benchmark
JitendraMalikCV · x · 2026-09-08
Robotics researcher Jitendra Malik pushes back on claims that LLMs like Astra have made serious robotics progress. He argues current demos show mostly simple pick-and-place tasks with parallel-jaw grippers, where LLMs excel at high-level planning — but robotics fundamentally requires dexterity and dynamics, demanding high-frequency control policies handling torques and forces.
His challenge: prompt an LLM to output high-frequency control commands for a legged robot traversing varied terrain — a task solved by 2021 RSS / 2022 CoRL work, i.e. five-year-old technology. He clarifies he doesn't doubt LLMs' value for planning and agentically assisting robotics research.
Related event: Robotics Legend Malik Pushes Back on LLM-Driven Robot Control(2 posts)→
More from Embodied
- Robotics researcher: harness + tool calls beat raw VLA on control benchmarks, at 2.3x lower cost — rbhar90 · 2026-09-08
- Grok bot plays a Roland piano, composing romantic-style music from a text prompt — yunta_tsai · 2026-09-08
- 360CityArena at ECCV 2026: best AI agent scores 17.1% vs humans' 77.3% on city navigation — CSProfKGD · 2026-09-08
- Open-source C-based LLM engine runs Gemma voice conversations on Jetson Orin, faster than llama.cpp — cortexist · 2026-09-08
- Self-driving demo runs on nothing but logic gates, moves like an insect — burny_tech · 2026-09-08
- Xpeng announces first automated production line where robots mass-produce robots — Anxious-Yoghurt-9207 · 2026-09-08