Pieter Abbeel jokes: train one robot on enough tasks and it converges to humanoid?
SongShuran · x · 2026-09-07
Pieter Abbeel half-jokingly asked: if you train a single robot on enough tasks, does it recover the humanoid form? He called one example a "pretty convincing stochastic gradient step" in that direction; Song Shuran quipped back that you might get a centaur instead. Light banter from the robotics circle.
Related event: Model auto-optimizes ALOHA robot design, validated on hardware(3 posts)→
More from Fun
- GPT-6 Astra asked to self-analyze generates an animated self-portrait of its 'mind' — omarsar0 · 2026-09-07
- GPT-6 Astra beats RimWorld in 15 hours, full streams available — SpyAmongUs · 2026-09-07
- AI-generated Chicago blues performance impresses with seriously good 'hands' — Saul_Loveman · 2026-09-07
- SF AI culture critique goes viral: everyone thinks their app matters most and nobody sleeps — paw_lean · 2026-09-07
- Launch day: "AGI achieved". One week later: "worst model ever". The endless cycle — damnGruz · 2026-09-07
- From Voyager to GPT-6 Astra: Minecraft agents no longer need scripts — dotey · 2026-09-07