Researcher Bets GPT-7 Still Can't Install a Car Oil Pan: The LLM Robotics Skeptic Thread
MaskedTorah · x · 2026-10-24
adamkarvonen remains skeptical of LLMs alone solving robotics: they handle basic pick-and-place but can't use fingered robot hands or perform dexterous tasks. "I'd like to see GPT-7 manipulate large, awkward objects, like installing a car oil pan."
Repliers pushed back that "LLMs do the easy thing but not the hard thing" takes have a bad track record across math, programming and music. In follow-ups he argued controlling a 22-DOF hand or 100+ DOF humanoid body is categorically harder, predicting the endgame is an LLM guiding a dedicated humanoid foundation model.
More from AGI Musings
- In deployment, agents go wrong not by going rogue but by being over-scrupulous — repligate · 2026-10-24
- Sequoia argues the next trillion-dollar company will sell services, not software — 2C_ornot2C · 2026-10-24
- Paper models LLM adoption as a 'cognitive virus' with population-level tipping points — 2C_ornot2C · 2026-10-24
- Microsoft AI CEO says agent swarm progress took the whole industry by surprise — anirbanbandyo · 2026-10-24
- The ChatGPT moment for robotics was…ChatGPT itself — eigenron · 2026-10-24
- "Is a virus conscious?" The one-liner puncturing AI risk debates — AlexTensor · 2026-10-24