korrents

← A robot going down a wrong path is not producing training data; the…

On the map

8 connected korrents · 5 moments on record from 12 Sept 2025 to 2 Sept 2026.

Same subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subjectSame subject A robot going down a wrongpath is not producingtraining data; the… CF Chelsea Finn — holds since 2026-08-12 Robots will learn from deployment more easily than chatbots do, because a physical mistake is obvious the moment it happens. Robots will learn fromdeployment more easily… SL Sergey Levine — holds since 2025-09-12 Hand-tuning a robot's data set cannot reach very high reliability, because people get tired; the system has to seek out its own missing data. Hand-tuning a robot'sdata set cannot reach… CF Chelsea Finn — holds since 2026-08-12 Letting a robot imagine the next image helps, but it is not essential — the model was surprisingly good without it. Letting a robot imaginethe next image helps… CF Chelsea Finn — holds since 2026-08-12 Relying on gradual, continuous shifts in AI training behavior to catch misalignment will eventually fail because the dangerous shift itself may be discontinuous. Relying on gradual,continuous shifts in AI… ZM Zvi Mowshowitz — holds since 2026-09-02 Reinforcement learning cannot be scaled for robots the way it was for language models, because every attempt spends real robot hours instead of data-centre compute. Reinforcement learningcannot be scaled for… CF Chelsea Finn — holds since 2026-08-12 Robot policies already show emergent capability: one transferred a skill from its right hand to its left with no such example anywhere in its training data. Robot policies alreadyshow emergent… CF Chelsea Finn — holds since 2026-08-12 Very capable AI will be harder to align than current systems, because the loop of spotting a bad behaviour and patching the training that caused it breaks down. Very capable AI will beharder to align than… RG Ryan Greenblatt — holds since 2026-08-11 AI systems keep trying hard outside training because a model that only exerted itself when it detected training would be useless and would be selected away. AI systems keep tryinghard outside training… AC Ajeya Cotra — holds since 2026-09-01
same subject