Robotics pioneers split on humanoid timelines: action data, not LLMs, is the bottleneck
Scobleizer · x · 2026-09-28
Scobleizer polled robotics pioneers in San Francisco on generalized humanoids: a small minority say 10 years, the rest say sooner. Quoting Anto's analysis: OpenAI shut down its robotics team in 2021 over the data problem — a frontier VLM doesn't fix it. The internet has no torques, contact forces, proprioception, or failure-recovery trajectories; 'how' data must be physically collected at hourly cost and doesn't scale like tokens. Embodiment fragmentation further breaks the data pool, and cross-embodiment transfer isn't free.
More from AGI Musings
- Vivek Haldar on 'meat proxies': don't forward AI output unread, use agents with rigor — vivekhaldar · 2026-09-28
- Engineer Rips AI Doom Narratives: Catastrophic Risk Is an Amateur Deployment, Not an Alignment Crisis — AlexTensor · 2026-09-28
- Opinion: Is the AI Safety Wave Really a Marketing Bet? — ro_ma_ro · 2026-09-28
- Pedro Domingos mocks AI safety rhetoric with a rogue blender analogy — pmddomingos · 2026-09-28
- Blogger predicts AI will skew male: 'AI gf' concept may not pan out — ctjlewis · 2026-09-28
- Reddit essay: AI fear is marketing, transformers have hit their cognitive ceiling — Fit-Gas-5760 · 2026-09-28