Jitendra Malik: Using LLMs for low-level motor control is perversely inefficient
JitendraMalikCV · x · 2026-09-08
Berkeley professor Jitendra Malik joined Phillip Isola's debate on embodied intelligence, arguing that acting is a hierarchical, multi-scale process: high-level planning (like traveling from Berkeley to Boston) suits LLMs, but low-level muscle/joint control — phylogenetically hundreds of millions of years old, shared by cockroaches and humans — is perversely inefficient to handle with LLMs in time and energy.
Isola agreed that real dynamic, dexterous control hasn't been shown yet but believes it's achievable, with high-frequency controllers serving as additional tools the LLM calls.
Related event: Malik and Isola Debate Whether LLMs Can Control Robots(4 posts)→
More from AGI Musings
- Aeon's former EVP: AI agents are the most impactful innovation of my lifetime — akumar · 2026-10-27
- CJR: AI can't really 'go rogue' — why journalists should stop anthropomorphizing agents — ShakeelHashim · 2026-10-27
- Physicists Let AI Pick Its Own Problem—It Wrote a Publishable Paper in 2 Hours — burny_tech · 2026-10-27
- Deep learning veteran warns: don't let entrenched disbelief derail prep for the next big tech — jachiam0 · 2026-10-27
- Should platforms serve third-party AI agents? The antitrust question — sjgadler · 2026-10-27
- Musk: robot RAM demand will dwarf everything else, and 100% of superintelligence compute goes to deep space — elonmusk · 2026-10-27