Three insights from a deep thinker with real experience. #2 is what I call “Video Emptor”: be wary of robot videos. Now they are often labeled “autonomous 4x speed up” but no indication of cherry-ness / robustness / success rate.
@ken_goldberg
-
AI as a Humbling Force: Moving Beyond Human Exceptionalism
By
–
The Copernican revolution taught us that we are not the center of the physical universe. AI is teaching us that we're not the only thinking beings. Can AI make us less arrogant and self-centered? One can hope.
-
Questions on Demo Statistics, Failure Modes, and Bimanual Grasping Correlations
By
–
Kudos on this elegant demo! Can you share some stats and failure modes? Are any of these items in the training set? Bimanual grasping seems to correlate w/ size. Are there counterexamples? Eg, a full-sized paper grocery bag w/ handle that it grasps w/ one hand? https://t.co/nITsKLCgG1
— Ken Goldberg (@Ken_Goldberg) 24 décembre 2025Kudos on this elegant demo! Can you share some stats and failure modes? Are any of these items in the training set? Bimanual grasping seems to correlate w/ size. Are there counterexamples? Eg, a full-sized paper grocery bag w/ handle that it grasps w/ one hand?
-
Robot Manipulation Discussion: Elon Musk, Commercialization and Real-World Data
By
–
Fun discussion about robot manipulation, @elonmusk (15 mins in), and ideas for making progress on getting specific robot tasks to be reliable and efficient so they can be commercialized and generate real world data.
-
Datasets: Accounting for Size and Quality Metrics
By
–
Is this # datasets without accounting for size or quality? https://t.co/7dv2Lv6EBj
— Ken Goldberg (@Ken_Goldberg) 24 décembre 2025Is this # datasets without accounting for size or quality?
-
AI and Physical Intelligence: A Historical Perspective on Embodiment
By
–
The distinction between AI and physical intelligence is now widely recognized, but that wasn't always the case. https://
withoutwhy.substack.com/p/ai-embodimen
t-and-the-limits-of-simulation
… @BScotRousse -
Multi-Fingered Hand Assembly: Engineering vs Machine Learning Trade-offs
By
–
Precision assembly is common but OK! Doing it with a multi-fingered hand is interesting: how much of this demo is due to Model-Free Learning and how much is Good Old-Fashioned Engineering (GOFE)? I'll bet it's mostly the latter. Now use it to generate data for the former…
-
New Toolkit for Extending Training Data to Different Robot Embodiments
By
–
Excited to release this toolkit for extending training data to different robot embodiments: https://t.co/sqhKr7akE8
— Ken Goldberg (@Ken_Goldberg) 17 décembre 2025Excited to release this toolkit for extending training data to different robot embodiments:
-
Brilliant Roboticist Teams Building Efficient Systems
By
–
Excellemt summary Lukas. Like @DynaRobotics and @SundayRobotics, @AmbiRobotics is led by a brilliant team of roboticists and engineers laser-focused on building systems that work efficiently in the real world to capture avalanches of real robot data for learning. https://t.co/NvUWdaNWWI
— Ken Goldberg (@Ken_Goldberg) 10 décembre 2025Excellent summary Lukas. Like @DynaRobotics and @SundayRobotics, @AmbiRobotics is led by a brilliant team of roboticists and engineers laser-focused on building systems that work efficiently in the real world to capture avalanches of real robot data for learning.
-
Waymo’s Use of Large Models and GOFE in Autonomous Driving
By
–
My sense is that @Waymo has been using large models for awhile (plus lots of GOFE for sensor pre-processing, system bootstrapping, data avalanching, and ongoing safety). One clue is the “thinking fast and slow” note on the image.