Steven Henderson

Beginning to think LLM improvement is asymptotic. Never better than the smartest, best informed, and persistent humans. Meaning recursive self learning will take us from accelerating to decelerating improvements.

@TheLtColUSMC Yeah just wear a working uniform? Or don’t stand there?

🚨 Hot take: Let’s use our chinstraps.

@adcock_brett Good to see no safety chains holding it up

F.03 can now climb a ladder, fully autonomous

@LeoKharon One day that robot is going to become self aware and…

New framework: Kick down your robot, it will get back up every time 🥋 Chinese startup RoboParty is a Beijing startup founded April 2025 by Huang Yi, originally shipping ROBOTO ORIGIN, the world's first full-stack open-source bipedal humanoid. They released UFO: Unsupervised Reinforcement Learning Framework for Humanoid Control. DEFINITIONS -> what differs is where the learning signal comes from: - SUPERVISED: humans supply the right answers (labels), the model imitates them. - UNSUPERVISED: no answer key, the model finds structure in raw data on its own. - REINFORCEMENT LEARNING: no answer key either, the model tries things and a reward scores each attempt. → UNSUPERVISED RL: trial and error where the agent invents its own rewards, instead of engineers hand-writing one per task. REPRESENTATION LEARNING: compress raw states into a useful internal map. TEMPORAL DISTANCE: distance on that map is "how many steps from A to B." CONTRASTIVE: trained by pulling together what's close in time, pushing apart what isn't. -> CONTRASTIVE TEMPORAL-DISTANCE REPRESENTATION LEARNING: the model builds an internal map of body states where distance means how many steps it takes to get from one to another. It is trained by contrast: states that occur close together in a movement get pulled together in the map, randomly paired states get pushed apart. UFO is an open-source training framework that teaches humanoid robots skills, like getting up, walking, goal-reaching, teleoperation, without reference motions -> no motion-capture or human-video demonstrations to imitate. Its core is TeCH, a contrastive temporal-distance representation-learning algorithm: the robot explores, builds pseudo-goals by temporal rolling, and learns goal-conditioned policies from a single unified progress reward. One framework trains five different robots (Unitree G1/H1, RoboParty RP0/RP1, AgiBot X2) with automatic config conversion in ~2–3 hours per robot! The real novelty here "no demonstrations at all". No data-collection arms race,the dominant humanoid-locomotion recipe is tracking: imitate mocap/retargeted-human reference trajectories. The robot self-generates goals from its own exploration and learns from a progress reward, needing zero reference motion data. Everybody else is fighting over data acquisition, while this team just teleports out of the race entirely (inb4 "competition is for losers 💀 ). This strategy reminds me of the DeepSeek playbook applied to robots: open-source the whole stack to become the global default and commoditize everyone else. RoboParty is giving away hardware and now control software (UFO) to be the Android of humanoids. Yet another reason for the US to ban Chinese open models perhaps 🥶 ? What I also really like about this approach is the cross-embodiment infrastructure, one framework trains Unitree G1/H1, RoboParty RP0/RP1, and AgiBot X2 with automatic configuration conversion. Just like Physical Intelligence, RoboParty seems to place itself as a neutral hardware agnostic middle man. Also woth mentioning: their ability ot perform stable skill injection, e.g. adding a cartwheel without forgetting how to walk. A common failure of RL humanoid policies is that teaching a new agile skill destabilizes the existing ones (catastrophic forgetting). UFO claims you can inject rare motions (cartwheel) without collapsing learned behavior. If it holds, that's a significant incremental/continual skill-learning! But again, I have to underline it: no arXiv, no external validation, no success-rate numbers. -> robotics badely needs an independent unbiased evaluator imho. Still, look at that cool demo: robot is getting kicked and pushed around (serious disturbance) during teleoperation (controlled the person at the back wearing the VR headset), and still managed to always get back up. This is some serious demonstration of stability and robustness!

@HiddenHistoryYT Highly skewed propellers that took generations and billions in R&D to make. A lady should cover up for a photo even if all the bad guys know what’s under the towel

USS John Paul Jones (DDG 53) Arleigh Burke-Class flight guided missile destroyer in drydock at Vigor Shipyard in Seattle, Washington.

@Defence_IDA Glad to see the PRC spend billions on building an almost equal weapon system that may eventually made irrelevant with real time satellites targeting and one-way drones

Alleged new footage shows the PLAN's J-35 and KJ-600 taking off and landing aboard the aircraft carrier Fujian.

@HiddenHistoryYT Back when men were men

Chicopee (AO 34) refueling Intrepid (CV 11) to port while simultaneously refueling Norman Scott (DD 690) to starboard, Pacific Ocean, 10 April 1945

@Rainmaker1973 Gives a preview of what meta-material camouflage will look like

Julian Voss-Andreae is a quantum physicist-turned-sculptor. His work is heavily influenced by his background in science and his blending figurative sculptures can vanish in front of our eyes.

@johnkonrad We need more people who keep friends from across the aisle. Takes true comfort in your own beliefs to have liberal friends and family if you are conservative and guessing vice versa

When I brought up Mamdani a liberal friend tried to convince me that Trump is actually a communist 🫠🫠🫠🫠🫠

@oscarmoxon Why so many antennae …?

so… this is real

@mercoglianos If they need help getting their spaceship back they should call @FleetzeroHQ

Go Australus, Normand Ranger and Skimmer Tide are still drifting westward. Not clear if any tow has been established yet with Starship.

@PrecisionLabs0 @grok based on this video is the SU-35 armed with air to air missiles?

A routine NORAD intercept took a dangerous turn in seconds. On September 23, 2024, cockpit footage captured a Russian Su-35 performing an aggressive close pass in front of a U.S. Air Force F-16 during a NORAD intercept of a Tu-95 bomber in the Alaskan ADIZ. NORAD later called the maneuver unsafe and unprofessional, warning that it unnecessarily endangered both aircraft and their crews. How close is too close during an aerial intercept?

@Not_the_Bee Bro is lucky he didn’t get mistaken for the shooter. Friendly fire is real

Unknown cowboy engages active shooter at Idaho In-N-Out notthebee.com/article/unknow…

@Amena__Bakr Wasn’t in the mood for tacos. Just wanted lasting peace in the Middle East and no psychos chanting for my demise

Happy taco day

@ApoStructura Thanks for this. The scale is incomprehensible

When people see visualizations of satellites in orbit they get scared because it looks crowded. So here is a visualization of what it looks like with accurate satellite sizes:

@bscholl @ApoStructura Would love to see an honest cost-benefit analysis of ISS

@ApoStructura Half an ISS long and a million times more useful

This thing is half an ISS long

@typesfast @levie Hovel is a good one

@levie How long before the peasant in his hovel benefited from Newton’s breakthroughs?

We’re going to see an increasing divergence between what AI does in our personal lives and in daily productivity vs. what it can do in very deep domains like math, science, legal, coding and more. Up to some threshold, capability was evenly felt across all domains because the models were just becoming mildly useful in general. Now, the deep domain work is about to go vertical. Most people won’t actually notice these benefits in their day to day life directly (indirectly they certainly will over time), but the experts in these fields will. And there’s no inherent ceiling to what capabilities are needed, unlike in the consumer space where needs can get met relatively straightforwardly. This will often lead to a capability overhang, though, as many of these performance gains need to be applied to data sets and workflows in an applied way for that area of work. But this is ultimately how you get breakthroughs in life sciences, real world automation, new cyber capabilities, and more.

A parallel to the 2nd law of thermodynamics or something

Beginning to think LLM improvement is asymptotic. Never better than the smartest, best informed, and persistent humans. Meaning recursive self learning will take us from accelerating to decelerating improvements.

@bscholl @sentdefender WWIII

@sentdefender Seems like Ukraine, Iran, Russia, China etc are all one war. 🤔

Russia is providing valuable satellite and signals intelligence to Iran that is likely helping it to pinpoint U.S. military forces across the Middle East as well as jam American-made aircraft and munitions used in strikes against Iran, U.S. and European officials with knowledge of the matter have told NBC News.

@typesfast @AlexanderPayton That goes to Houston

@AlexanderPayton They can stop in LA on the way to Houston they just can’t load any cargo in LA