Video yükleniyor...

Video Yüklenemedi

Ana Sayfaya Dön

MIT’s latest research just just gave Robots a great new skill to grip delicate yet heavy objects without breaking them. The big problem is robots often cannot pick up heavy but fragile things because gripping hard enough usually means crushing or slipping. A single gripper shape has to do...

rohanpaul_ai's profile picture

Rohan Paul

146,828 subscribers

1,632,855 görüntüleme • 7 ay önce •via X (Twitter)

EğitimSağlık & İyilikBilim & Teknoloji
LIVE

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

Robots still fail at the simplest job: picking stuff up. 🫴🏼 Robots are good at repetition, but most still fail at grasping varied objects. In warehouses and grocery fulfillment, that’s the bottleneck: items aren’t uniform, and rigid grippers break or drop them. Nexera-Robotics built NeuraGrasp to address this. The system combines a compliant surface that mimics human fingertips with an AI controller that adjusts grip in real time. It doesn’t need reprogramming for each SKU and can handle fragile, irregular items as well as standard packages. This matters because if grasping is solved, robots stop being limited to narrow workflows. They can finally automate the “last meter” of logistics and retail, where most of the labor costs remain today. It's oddly satisfying to see a gripper that can pick up (almost) anything! 🫶🏼
0:53

Robots still fail at the simplest job: picking stuff up. 🫴🏼 Robots are good at repetition, but most still fail at grasping varied objects. In warehouses and grocery fulfillment, that’s the bottleneck: items aren’t uniform, and rigid grippers break or drop them. Nexera-Robotics built NeuraGrasp to address this. The system combines a compliant surface that mimics human fingertips with an AI controller that adjusts grip in real time. It doesn’t need reprogramming for each SKU and can handle fragile, irregular items as well as standard packages. This matters because if grasping is solved, robots stop being limited to narrow workflows. They can finally automate the “last meter” of logistics and retail, where most of the labor costs remain today. It's oddly satisfying to see a gripper that can pick up (almost) anything! 🫶🏼

Lukas Ziegler

24,055 görüntüleme • 11 ay önce

Why did so many languages copy async/await from C#? Anders Hejlsberg(Anders Hejlsberg) - creator of TypeScript, C# & Turbo Pascal - on what they got right with the design: #1 - async/await was designed to solve a common problem in the event-loop model: "A lot of languages are built around cooperative multitasking in the sense that they have an event loop that sits and dispatches events. Then you handle the event and then you yield back to the event handler loop. And it all runs in a single thread cooperatively. The problem with that is if you then want to do some long running work: how do I stop in the middle of this piece of long running work and yield back to the event loop cooperatively? And then when my result is ready, I can come back and continue executing here." #2 - state machines are the solution, but hard to build: "Well, in order to do that in an inverted architecture like that, you have to build a state machine. State machines are notoriously hard for people to implement because you've got to move all of your state off of the stack into objects. And then you have this big case statement that envelopes your entire logic. It's a nightmare to figure out. But, the transformation from serially executing code into a state machine, its continuation-passing-style translation is actually one that you can do in a machine-based fashion." #3 - compilers are good at writing state machines: "You can have the compiler write the state machine if you introduce syntax that allows you to indicate where you want to yield. And that's what await is. Await is basically saying, I want to yield here, and I want to yield this promise, and then when the promise completes, I want you to come back here and continue executing. Then the compiler writes a state machine around it and it actually turns it into this big switch statement and moves all of the state that survives across the await into something that's heap allocated. So it can be brought back. And doing all of that work is something that compilers are great at. And so that was sort of the idea that we have this new style of programming where we're using promises or the equivalent of promises and the ability to yield and then we have callbacks. But trying to write your program in that style, that's also what JavaScript suffered from a lot. It's like all this callback style stuff. With Async and Await, you get the illusion that you're just writing normal sequential code and then the compiler does the painful transformation for you. That turns out to be really useful."
3:13

Why did so many languages copy async/await from C#? Anders Hejlsberg(Anders Hejlsberg) - creator of TypeScript, C# & Turbo Pascal - on what they got right with the design: #1 - async/await was designed to solve a common problem in the event-loop model: "A lot of languages are built around cooperative multitasking in the sense that they have an event loop that sits and dispatches events. Then you handle the event and then you yield back to the event handler loop. And it all runs in a single thread cooperatively. The problem with that is if you then want to do some long running work: how do I stop in the middle of this piece of long running work and yield back to the event loop cooperatively? And then when my result is ready, I can come back and continue executing here." #2 - state machines are the solution, but hard to build: "Well, in order to do that in an inverted architecture like that, you have to build a state machine. State machines are notoriously hard for people to implement because you've got to move all of your state off of the stack into objects. And then you have this big case statement that envelopes your entire logic. It's a nightmare to figure out. But, the transformation from serially executing code into a state machine, its continuation-passing-style translation is actually one that you can do in a machine-based fashion." #3 - compilers are good at writing state machines: "You can have the compiler write the state machine if you introduce syntax that allows you to indicate where you want to yield. And that's what await is. Await is basically saying, I want to yield here, and I want to yield this promise, and then when the promise completes, I want you to come back here and continue executing. Then the compiler writes a state machine around it and it actually turns it into this big switch statement and moves all of the state that survives across the await into something that's heap allocated. So it can be brought back. And doing all of that work is something that compilers are great at. And so that was sort of the idea that we have this new style of programming where we're using promises or the equivalent of promises and the ability to yield and then we have callbacks. But trying to write your program in that style, that's also what JavaScript suffered from a lot. It's like all this callback style stuff. With Async and Await, you get the illusion that you're just writing normal sequential code and then the compiler does the painful transformation for you. That turns out to be really useful."

The Pragmatic Engineer

13,258 görüntüleme • 1 ay önce

🇺🇸 MIT and Stanford engineers built a robotic gripper that works like a vine. And it might be one of the more quietly important tech stories of the year. Soft inflatable tubes grow outward, twist and coil around an object, then retract, creating a gentle hammock-like cradle. It can handle a glass vase. It can handle a watermelon. A larger version can lift a full human body out of bed without injury. The immediate target application: eldercare. Lifting a bedridden patient is one of the most physically damaging tasks a caregiver performs, for both the caregiver and the patient. Beyond that: warehouse sorting, agricultural harvesting, heavy cargo loading... anywhere you need something strong enough to lift serious weight but gentle enough not to crush what it's holding. Most robotics tries to replicate the human hand. These researchers copied a plant instead. That's the idea worth paying attention to. Source: Robotics & Automation News, Interesting Engineering
0:25

🇺🇸 MIT and Stanford engineers built a robotic gripper that works like a vine. And it might be one of the more quietly important tech stories of the year. Soft inflatable tubes grow outward, twist and coil around an object, then retract, creating a gentle hammock-like cradle. It can handle a glass vase. It can handle a watermelon. A larger version can lift a full human body out of bed without injury. The immediate target application: eldercare. Lifting a bedridden patient is one of the most physically damaging tasks a caregiver performs, for both the caregiver and the patient. Beyond that: warehouse sorting, agricultural harvesting, heavy cargo loading... anywhere you need something strong enough to lift serious weight but gentle enough not to crush what it's holding. Most robotics tries to replicate the human hand. These researchers copied a plant instead. That's the idea worth paying attention to. Source: Robotics & Automation News, Interesting Engineering

Mario Nawfal

263,648 görüntüleme • 2 ay önce

i watched gemma 4 12b build something genuinely impressive today, and then loop itself to death right in front of me. the full run is in the video, sped up but completely uncut, watch it to the end and you will catch the exact moment it stops building and starts looping right in the middle of the work. the task was clean, build a single file gravity simulator, n-body physics, orbits, collisions, running locally on one 3090 through an agent. and for ten minutes it was a joy to watch. it reached for a symplectic integrator on its own, the correct one, the kind that keeps orbits stable instead of spiralling out. real gravity with softening, proper orbital velocities, momentum conserved on collision. the physics was right. the thing actually worked. then on the very last step, writing a few tests to prove its own code, it fell into a loop. not a crash, a loop. it started repeating itself and would not stop. ten more minutes, thirty four thousand tokens into a single answer, the same fragments over and over, until i killed it myself. so it's not that gemma can't code. it did the hard part beautifully. it cannot finish. it cannot hold a long task together without unravelling, and finishing is the entire job in agentic work. here's the part that stings. i run this exact task, same harness, same card, on the chinese open models, qwen especially, and i never see this. they build it, they test it, they stop. every single time. google has the raw capability, you can see it sitting right there in the code, and then the model loops itself to death on a task a 27b from alibaba finishes clean. open weights, apache 2.0, so much to love on paper. i just need it to know when to stop talking.
4:15

i watched gemma 4 12b build something genuinely impressive today, and then loop itself to death right in front of me. the full run is in the video, sped up but completely uncut, watch it to the end and you will catch the exact moment it stops building and starts looping right in the middle of the work. the task was clean, build a single file gravity simulator, n-body physics, orbits, collisions, running locally on one 3090 through an agent. and for ten minutes it was a joy to watch. it reached for a symplectic integrator on its own, the correct one, the kind that keeps orbits stable instead of spiralling out. real gravity with softening, proper orbital velocities, momentum conserved on collision. the physics was right. the thing actually worked. then on the very last step, writing a few tests to prove its own code, it fell into a loop. not a crash, a loop. it started repeating itself and would not stop. ten more minutes, thirty four thousand tokens into a single answer, the same fragments over and over, until i killed it myself. so it's not that gemma can't code. it did the hard part beautifully. it cannot finish. it cannot hold a long task together without unravelling, and finishing is the entire job in agentic work. here's the part that stings. i run this exact task, same harness, same card, on the chinese open models, qwen especially, and i never see this. they build it, they test it, they stop. every single time. google has the raw capability, you can see it sitting right there in the code, and then the model loops itself to death on a task a 27b from alibaba finishes clean. open weights, apache 2.0, so much to love on paper. i just need it to know when to stop talking.

Sudo su

39,574 görüntüleme • 2 ay önce

This work makes a humanoid robot do simple parkour moves by looking with a depth camera and choosing the right move on the fly. The big deal is that it turns lots of small human moves into long, real-time robot behavior, without hand-coding every transition or retraining for each new course. A humanoid robot is usually good at steady walking, but it often fails when it has to do fast moves like jumping up, vaulting, or rolling, and then keep going to the next obstacle. The hard part is that you cannot easily collect training data for every possible obstacle shape, distance, and mistake, so robots end up learning a few moves that only work in a narrow setup. This work starts from short clips of real human parkour moves, like stepping over, vaulting, climbing, and rolling. It uses motion matching, which is basically a smart “pick the next clip that fits best right now” search, to stitch those short clips into a long, smooth plan that looks like a human doing a whole course. Then it trains a controller with reinforcement learning (RL), which means the robot learns by trial and error to copy that plan while staying balanced and not falling. After training separate expert controllers for different moves, it compresses them into 1 controller that uses only onboard depth sensing and a simple “go this fast in this direction” command. In real tests on a Unitree G1 humanoid, it can clear multiple obstacles in a row, adapt when obstacles get moved, and climb a wall up to 1.25m.
0:05

This work makes a humanoid robot do simple parkour moves by looking with a depth camera and choosing the right move on the fly. The big deal is that it turns lots of small human moves into long, real-time robot behavior, without hand-coding every transition or retraining for each new course. A humanoid robot is usually good at steady walking, but it often fails when it has to do fast moves like jumping up, vaulting, or rolling, and then keep going to the next obstacle. The hard part is that you cannot easily collect training data for every possible obstacle shape, distance, and mistake, so robots end up learning a few moves that only work in a narrow setup. This work starts from short clips of real human parkour moves, like stepping over, vaulting, climbing, and rolling. It uses motion matching, which is basically a smart “pick the next clip that fits best right now” search, to stitch those short clips into a long, smooth plan that looks like a human doing a whole course. Then it trains a controller with reinforcement learning (RL), which means the robot learns by trial and error to copy that plan while staying balanced and not falling. After training separate expert controllers for different moves, it compresses them into 1 controller that uses only onboard depth sensing and a simple “go this fast in this direction” command. In real tests on a Unitree G1 humanoid, it can clear multiple obstacles in a row, adapt when obstacles get moved, and climb a wall up to 1.25m.

Rohan Paul

37,121 görüntüleme • 5 ay önce

Dario Amodei just told software engineers exactly how long they have. Six to twelve months. Amodei: “I have engineers within Anthropic who say I don’t write any code anymore. I just let the model write the code, I edit it, I do the things around it.” The people building the most powerful AI in history have already stopped writing code. That is not a forecast. That is the current working condition inside the lab closest to the frontier. Amodei: “We might be six to 12 months away from when the model is doing most, maybe all, of what SWEs do end-to-end.” The tech industry spent a decade making software engineers its highest-paid, most protected class. That era has a last day now. When a model can execute an entire software build end-to-end, the ability to write syntax stops being a skill. It becomes a credential for a job that no longer exists. Amodei: “And then it’s a question of how fast does that loop close.” That is the sentence everyone skipped. The code was never the hard part. The hard part was everything around it. The model just learned everything around it. Writing the code is already nearly gone. Testing is next. Deployment is next. When all three collapse into a single autonomous execution loop, the machine no longer needs a human in the chain at all. The corporation or sovereign state that closes that loop first does not gain a competitive advantage. It gains a category of speed that biological engineers cannot match, track, or reverse. That is not disruption. That is replacement at a systems level. Amodei is not describing a future disruption. He is describing the current state of his own building. The loop is already closing. The only question is whether you are inside it or outside it when it seals.
0:27

Dario Amodei just told software engineers exactly how long they have. Six to twelve months. Amodei: “I have engineers within Anthropic who say I don’t write any code anymore. I just let the model write the code, I edit it, I do the things around it.” The people building the most powerful AI in history have already stopped writing code. That is not a forecast. That is the current working condition inside the lab closest to the frontier. Amodei: “We might be six to 12 months away from when the model is doing most, maybe all, of what SWEs do end-to-end.” The tech industry spent a decade making software engineers its highest-paid, most protected class. That era has a last day now. When a model can execute an entire software build end-to-end, the ability to write syntax stops being a skill. It becomes a credential for a job that no longer exists. Amodei: “And then it’s a question of how fast does that loop close.” That is the sentence everyone skipped. The code was never the hard part. The hard part was everything around it. The model just learned everything around it. Writing the code is already nearly gone. Testing is next. Deployment is next. When all three collapse into a single autonomous execution loop, the machine no longer needs a human in the chain at all. The corporation or sovereign state that closes that loop first does not gain a competitive advantage. It gains a category of speed that biological engineers cannot match, track, or reverse. That is not disruption. That is replacement at a systems level. Amodei is not describing a future disruption. He is describing the current state of his own building. The loop is already closing. The only question is whether you are inside it or outside it when it seals.

Dustin

318,457 görüntüleme • 5 ay önce

Researchers at Netflix just released a new AI model It erases objects from video, then rewrites the physics of the entire scene as if that object never existed It's called VOID (Video Object and Interaction Deletion) Current inpainting tools simply paint over the gap left by an object's removal, but they ignore the chain of events that would inevitably follow. For example: > If you remove a car from a crash scene with a basic AI tool, and the other car still looks like it collides with something invisible > But VOID understands that once the car is gone, the other one would keep moving forward without impact The output is still rough and limited to a few seconds of footage, but with Netflix's entire content library to train on... this tech has a very clear path to improve.
0:38

Researchers at Netflix just released a new AI model It erases objects from video, then rewrites the physics of the entire scene as if that object never existed It's called VOID (Video Object and Interaction Deletion) Current inpainting tools simply paint over the gap left by an object's removal, but they ignore the chain of events that would inevitably follow. For example: > If you remove a car from a crash scene with a basic AI tool, and the other car still looks like it collides with something invisible > But VOID understands that once the car is gone, the other one would keep moving forward without impact The output is still rough and limited to a few seconds of footage, but with Netflix's entire content library to train on... this tech has a very clear path to improve.

Rowan Cheung

46,982 görüntüleme • 4 ay önce

A TURING AWARD WINNER STOOD UP AND TOLD A ROOM OF ENGINEERS THAT ALMOST NONE OF THEM DO THE ONE THING THAT ACTUALLY SEPARATES REAL ENGINEERING FROM TYPING, THEY WRITE CODE BUT THEY NEVER WRITE THE BLUEPRINT 56 minutes from Leslie Lamport -- Turing Award winner, creator of the tech behind almost every distributed system on earth. -> His claim: architects draw plans before a brick is laid. Programmers just start coding and hope. That gap is where complex systems quietly break. 04:10 -- A blueprint for software is called a spec. Write what the system must DO before touching how it does it. 10:13 -- Thinking above the code is the skill. The language is just the last, easiest step. 43:23 -- What programmers should really know isn't syntax. It's thinking clearly enough that "done" actually means done. 48:15 -- Thinking is hard, so we skip it. We jump straight to code because typing feels like progress. And this is exactly the wall the new "Loop engineering" hype is about to hit -- people now design loops of agents that write and check code while they walk away. But an unattended loop with no spec just ships broken work faster. Lamport is the missing half: the discipline that makes it safe to leave a loop running at all. You thought the leverage moved into better prompts and better loops. This is the man showing it moved above the code, where almost no one is willing to think. Save this. Read it before you trust a single agent loop ↓
56:47

A TURING AWARD WINNER STOOD UP AND TOLD A ROOM OF ENGINEERS THAT ALMOST NONE OF THEM DO THE ONE THING THAT ACTUALLY SEPARATES REAL ENGINEERING FROM TYPING, THEY WRITE CODE BUT THEY NEVER WRITE THE BLUEPRINT 56 minutes from Leslie Lamport -- Turing Award winner, creator of the tech behind almost every distributed system on earth. -> His claim: architects draw plans before a brick is laid. Programmers just start coding and hope. That gap is where complex systems quietly break. 04:10 -- A blueprint for software is called a spec. Write what the system must DO before touching how it does it. 10:13 -- Thinking above the code is the skill. The language is just the last, easiest step. 43:23 -- What programmers should really know isn't syntax. It's thinking clearly enough that "done" actually means done. 48:15 -- Thinking is hard, so we skip it. We jump straight to code because typing feels like progress. And this is exactly the wall the new "Loop engineering" hype is about to hit -- people now design loops of agents that write and check code while they walk away. But an unattended loop with no spec just ships broken work faster. Lamport is the missing half: the discipline that makes it safe to leave a loop running at all. You thought the leverage moved into better prompts and better loops. This is the man showing it moved above the code, where almost no one is willing to think. Save this. Read it before you trust a single agent loop ↓

slash1s

15,306 görüntüleme • 21 gün önce

since everyone woke up talking about loops, this article talks about domain chips, which i think are super underrated compared to loops.. a chip is a small specialist that's only good at one job QA, trading, content, research whatever ... and not one giant brain trying to do everything.. you wrap it in three things: >a workflow (same shape every time, so it's stable) >the tools that job needs >a benchmark that says what good looks what makes this special is the benchmark; without it, a loop just changes things randomly, when with it, the chip can score every pass and keep only what actually wins.. the problem with loops is they can spin 1000× and still not improve when a domain chip not only focuses on the domain you delegate to them, but they actually improve themselves with every spin. you can start with domain chips here this repo will help you build your first chip, the workflow, the benchmark pack, and the autoloop policy, with guardrails so it can't fool itself: below you can read the article fully + i put a good example of the difference between having a normal loop and a domain chip..
0:25

since everyone woke up talking about loops, this article talks about domain chips, which i think are super underrated compared to loops.. a chip is a small specialist that's only good at one job QA, trading, content, research whatever ... and not one giant brain trying to do everything.. you wrap it in three things: >a workflow (same shape every time, so it's stable) >the tools that job needs >a benchmark that says what good looks what makes this special is the benchmark; without it, a loop just changes things randomly, when with it, the chip can score every pass and keep only what actually wins.. the problem with loops is they can spin 1000× and still not improve when a domain chip not only focuses on the domain you delegate to them, but they actually improve themselves with every spin. you can start with domain chips here this repo will help you build your first chip, the workflow, the benchmark pack, and the autoloop policy, with guardrails so it can't fool itself: below you can read the article fully + i put a good example of the difference between having a normal loop and a domain chip..

JUMPERZ

36,537 görüntüleme • 2 ay önce

.David Deutsch: "What's currently called AI and AGI are not only different from each other, they are very close to being the exact opposites of each other. The reason is that an AI, current AI is like an AI that diagnoses diseases or an AI that plays chess or an AI that controls a huge factory. Those things have objective functions, that is they have a function that they are designed to maximize and that is why they are used in those particular applications. Or in military terms, you could say the objective is to hit the target. You might say the objective is to hit the target unless some thing specified, but it's a specified thing comes up in which case don't hit the target and so on. This is, as I said, almost the opposite of what humans do when humans think. For a start, the AI has to be obedient, that is it has to actually do the things it is programmed to do, whereas a human is fundamentally disobedient, especially when being creative. When a human plays chess, they are performing a completely different kind of computation. They don't do the same things, they don't investigate the same possibilities that the artificial chess playing machine does, because the artificial one is capable of looking at billions and billions of possibilities, whereas the human can only look at hundreds or something. They are doing something completely different. Another difference is that the human can explain, can write a book later, having become world champion, can write a book saying how I did it, as the computer program that beats the world champion can write no such book, because it has no idea how it did it. It was just following a program. I was doing this and that and that and none of that is illuminating. Also, third thing, the chess player can decide I don't want to play chess anymore, from now on I will play Go or from now on I will play tennis. If commanded to play chess, the functionality will deteriorate completely. Those things are different. What we want in an AGI is that it behaves in a way that cannot be specified in advance, because if you specified it, you would already have the answer. The AGI program has to give unexpected answers, answers to questions we didn't even know how to ask."
3:11

.David Deutsch: "What's currently called AI and AGI are not only different from each other, they are very close to being the exact opposites of each other. The reason is that an AI, current AI is like an AI that diagnoses diseases or an AI that plays chess or an AI that controls a huge factory. Those things have objective functions, that is they have a function that they are designed to maximize and that is why they are used in those particular applications. Or in military terms, you could say the objective is to hit the target. You might say the objective is to hit the target unless some thing specified, but it's a specified thing comes up in which case don't hit the target and so on. This is, as I said, almost the opposite of what humans do when humans think. For a start, the AI has to be obedient, that is it has to actually do the things it is programmed to do, whereas a human is fundamentally disobedient, especially when being creative. When a human plays chess, they are performing a completely different kind of computation. They don't do the same things, they don't investigate the same possibilities that the artificial chess playing machine does, because the artificial one is capable of looking at billions and billions of possibilities, whereas the human can only look at hundreds or something. They are doing something completely different. Another difference is that the human can explain, can write a book later, having become world champion, can write a book saying how I did it, as the computer program that beats the world champion can write no such book, because it has no idea how it did it. It was just following a program. I was doing this and that and that and none of that is illuminating. Also, third thing, the chess player can decide I don't want to play chess anymore, from now on I will play Go or from now on I will play tennis. If commanded to play chess, the functionality will deteriorate completely. Those things are different. What we want in an AGI is that it behaves in a way that cannot be specified in advance, because if you specified it, you would already have the answer. The AGI program has to give unexpected answers, answers to questions we didn't even know how to ask."

Deutsch Explains

72,455 görüntüleme • 1 yıl önce

Dr Fei-Fei-Li explains with a simple example how everyday household chores are so extremely difficult for Robots. "If you tell a robot to open the top drawer and watch out for the vase, this is actually a really hard task for robots." because the robot must ground language into the real world. Words like "top", "drawer", and "vase" are abstract. The system has to map them to 3D locations, objects, and relations in a noisy scene. This requires robust perception, object recognition, and spatial reasoning under uncertainty. The robot also lacks human commonsense. "Watch out" implies predicting consequences, estimating clearances, and understanding that vases are fragile. Encoding such priors, like how heavy a drawer is or how a vase might tip, is very complex and difficult without rich world knowledge. Learning the behavior from rewards is tough. The success signal is very sparse here, so naive exploration almost never stumbles on a full success sequence. This makes policy learning sample inefficient and brittle, especially when the environment changes between training and deployment. A sparse reward situation is when the agent only gets a success signal at the very end, and gets little or no feedback along the way. If a robot must open a drawer without hitting a vase, it might get reward only if the drawer ends up open and the vase is intact. Every partial try before that looks the same to the learner, reward equals 0. --- From "DSAI by Dr. Osbert Tay" YT channel
0:48

Dr Fei-Fei-Li explains with a simple example how everyday household chores are so extremely difficult for Robots. "If you tell a robot to open the top drawer and watch out for the vase, this is actually a really hard task for robots." because the robot must ground language into the real world. Words like "top", "drawer", and "vase" are abstract. The system has to map them to 3D locations, objects, and relations in a noisy scene. This requires robust perception, object recognition, and spatial reasoning under uncertainty. The robot also lacks human commonsense. "Watch out" implies predicting consequences, estimating clearances, and understanding that vases are fragile. Encoding such priors, like how heavy a drawer is or how a vase might tip, is very complex and difficult without rich world knowledge. Learning the behavior from rewards is tough. The success signal is very sparse here, so naive exploration almost never stumbles on a full success sequence. This makes policy learning sample inefficient and brittle, especially when the environment changes between training and deployment. A sparse reward situation is when the agent only gets a success signal at the very end, and gets little or no feedback along the way. If a robot must open a drawer without hitting a vase, it might get reward only if the drawer ends up open and the vase is intact. Every partial try before that looks the same to the learner, reward equals 0. --- From "DSAI by Dr. Osbert Tay" YT channel

Rohan Paul

342,627 görüntüleme • 8 ay önce

The yen just jumped hard after intervention and everyone is already relaxing like the problem is fixed. That’s the dangerous part. When the yen was getting crushed, everyone stayed alert. Positioning was careful. People were watching every move. Then Japan and the US stepped in, the yen jumped, and the tone flipped almost overnight. Relief showed up. People started talking like the pressure had been handled. That shift is the part that matters. Nothing about the real setup changed. The rate gap is still wide. You can still borrow yen cheap and put that money to work somewhere that pays more. That incentive did not disappear just because two governments defended the currency for a few days. The door is still open. All the intervention did was interrupt the move. It did not close it. Now we are in the quiet stretch. Headlines slow down. The yen stops making dramatic daily swings. And that is usually when people start leaning back into the same trades that created the pressure in the first place. Caution fades. Positioning rebuilds. The market starts testing whether the threat of more intervention is real or just something officials said while the cameras were on. I do not like this phase. A sharp selloff forces people to pay attention. A bounce lets them look away. And once enough people look away, the market does not need a brand new crisis. It only needs the existing pressure to reassert itself. That is why the current calm bothers me more than the original move did. Delayed pressure does not disappear. It waits. And when it stops waiting, it usually does not give much warning. You’re treating this bounce like the problem is solved… or are you watching the part that usually gets people hurt? #Gold $JPY #Kospi
5:58

The yen just jumped hard after intervention and everyone is already relaxing like the problem is fixed. That’s the dangerous part. When the yen was getting crushed, everyone stayed alert. Positioning was careful. People were watching every move. Then Japan and the US stepped in, the yen jumped, and the tone flipped almost overnight. Relief showed up. People started talking like the pressure had been handled. That shift is the part that matters. Nothing about the real setup changed. The rate gap is still wide. You can still borrow yen cheap and put that money to work somewhere that pays more. That incentive did not disappear just because two governments defended the currency for a few days. The door is still open. All the intervention did was interrupt the move. It did not close it. Now we are in the quiet stretch. Headlines slow down. The yen stops making dramatic daily swings. And that is usually when people start leaning back into the same trades that created the pressure in the first place. Caution fades. Positioning rebuilds. The market starts testing whether the threat of more intervention is real or just something officials said while the cameras were on. I do not like this phase. A sharp selloff forces people to pay attention. A bounce lets them look away. And once enough people look away, the market does not need a brand new crisis. It only needs the existing pressure to reassert itself. That is why the current calm bothers me more than the original move did. Delayed pressure does not disappear. It waits. And when it stops waiting, it usually does not give much warning. You’re treating this bounce like the problem is solved… or are you watching the part that usually gets people hurt? #Gold $JPY #Kospi

₿Strategy 🟧

16,945 görüntüleme • 3 gün önce

Opus 4.5 (new model from Anthropic) can do something no model could do until now: It can close the design loop. This is (in my testing) the first model that can reliably do the visual loop of Generate Code ⇛ Render ⇛ Look at it ⇛ Improve. Step 3 (Visual critique) has been a real problem for so long. Models will be myopic or defensive when reviewing designs - it's been hard to tell if this has been a training or a vision problem. Opus 4.5 seems to have fixed it - it's not amazing, but it finally *can* do it. Claude models have historically (once again, in my testing) lagged behind in terms of visual understanding. Gemini for the last year has been the best - but there might be a new leader.
10:06

Opus 4.5 (new model from Anthropic) can do something no model could do until now: It can close the design loop. This is (in my testing) the first model that can reliably do the visual loop of Generate Code ⇛ Render ⇛ Look at it ⇛ Improve. Step 3 (Visual critique) has been a real problem for so long. Models will be myopic or defensive when reviewing designs - it's been hard to tell if this has been a training or a vision problem. Opus 4.5 seems to have fixed it - it's not amazing, but it finally *can* do it. Claude models have historically (once again, in my testing) lagged behind in terms of visual understanding. Gemini for the last year has been the best - but there might be a new leader.

Hrishi

57,822 görüntüleme • 8 ay önce

This is by far my favourite minute of the $OPEN house. For anyone who doesn't understand what $OPEN wants to do, the CEO gives you a tell right here. This is happening in real time right now. This man is a genius. Basically $OPEN wants to own everything when it comes to buying and selling a home, the whole chain. As he says in the clip: "The buying, the selling, the mortgage and the insurance." So they can offer a seller a price he or she can't refuse (given perks like speed, certainty, and other soft attributes a traditional broker just cannot guarantee). They can offer the buyer a fair price, with the same perks. Speed, simplicity, certainty. Their AI mortgage gives the buyer the best price on the market and hooks the buyer for life. And then insurance on top of that. And there you go! A loyal and happy customer for life. As Kaz explains further. The easier this process becomes, the easier it is to move, more demand will be created. Thus more customers. And the loop continues. Now tell me, what company can compete with $OPEN when this is done? (and it is happening right now). Just the moat of $OPEN will be worth an insane amount of money.
1:07

This is by far my favourite minute of the $OPEN house. For anyone who doesn't understand what $OPEN wants to do, the CEO gives you a tell right here. This is happening in real time right now. This man is a genius. Basically $OPEN wants to own everything when it comes to buying and selling a home, the whole chain. As he says in the clip: "The buying, the selling, the mortgage and the insurance." So they can offer a seller a price he or she can't refuse (given perks like speed, certainty, and other soft attributes a traditional broker just cannot guarantee). They can offer the buyer a fair price, with the same perks. Speed, simplicity, certainty. Their AI mortgage gives the buyer the best price on the market and hooks the buyer for life. And then insurance on top of that. And there you go! A loyal and happy customer for life. As Kaz explains further. The easier this process becomes, the easier it is to move, more demand will be created. Thus more customers. And the loop continues. Now tell me, what company can compete with $OPEN when this is done? (and it is happening right now). Just the moat of $OPEN will be worth an insane amount of money.

Mr Never Sell

15,672 görüntüleme • 2 ay önce

Humanoids are finally getting a sense of touch that actually feels... human. 🤖 As third-gen bots like Optimus, XPENG, and Figure move from labs to the real world, the focus is shifting from just "walking" to "feeling." We are seeing a move toward tactile fabrics that can be literally tailored to a robot's frame like a custom suit. This isn't just about simple pressure sensors; it's about a full electronic skin with a folding radius under 0.2mm. That means a bot can detect a gram-level touch, sense textures, or even feel the slight slip of an object before it drops. The industrial impact is massive—the global market for these flexible sensors is projected to hit 27.4 billion RMB (approx. $3.98B) by 2030. Beyond just picking up boxes, this tech allows robots to sense the warmth and pressure of a hug or adjust balance through foot sensors. High-density arrays with dozens of sensors per square centimeter are becoming the new baseline for dexterity. By turning physical contact into high-fidelity data, these skins are the final key to making robots safe for our homes and hospitals. The future of robotics is looking a lot more flexible and a lot more sensitive.
0:25

Humanoids are finally getting a sense of touch that actually feels... human. 🤖 As third-gen bots like Optimus, XPENG, and Figure move from labs to the real world, the focus is shifting from just "walking" to "feeling." We are seeing a move toward tactile fabrics that can be literally tailored to a robot's frame like a custom suit. This isn't just about simple pressure sensors; it's about a full electronic skin with a folding radius under 0.2mm. That means a bot can detect a gram-level touch, sense textures, or even feel the slight slip of an object before it drops. The industrial impact is massive—the global market for these flexible sensors is projected to hit 27.4 billion RMB (approx. $3.98B) by 2030. Beyond just picking up boxes, this tech allows robots to sense the warmth and pressure of a hug or adjust balance through foot sensors. High-density arrays with dozens of sensors per square centimeter are becoming the new baseline for dexterity. By turning physical contact into high-fidelity data, these skins are the final key to making robots safe for our homes and hospitals. The future of robotics is looking a lot more flexible and a lot more sensitive.

RoboHub🤖

72,367 görüntüleme • 4 ay önce

AI + robotics research is starting to pick up steam. We can now instruct Spot using natural language using Language-guided Skill Coordination (LSC). A user provides a natural language instruction: "Bring me the chocolates box, cereal box, and pill bottle, and put them on the bedroom table" and the robot navigates to the location of the target objects and places them on the room table. Current versions of Spot rely on a fixed-set vocabulary that cannot generalize to diverse instructions. In this demo, the researchers presented a method that uses large language models to receive a free-form natural language instruction for object rearrangement, which it then executes. It combines: -A voice-to-text model that processes the instructions into text -An LLM that converts natural language instruction to call a library of skills -A perception module that provides ground truth locations and visual object detection And what your left with is a glimpse into what our future will hold: robots executing tasks from human voice prompts.
0:30

AI + robotics research is starting to pick up steam. We can now instruct Spot using natural language using Language-guided Skill Coordination (LSC). A user provides a natural language instruction: "Bring me the chocolates box, cereal box, and pill bottle, and put them on the bedroom table" and the robot navigates to the location of the target objects and places them on the room table. Current versions of Spot rely on a fixed-set vocabulary that cannot generalize to diverse instructions. In this demo, the researchers presented a method that uses large language models to receive a free-form natural language instruction for object rearrangement, which it then executes. It combines: -A voice-to-text model that processes the instructions into text -An LLM that converts natural language instruction to call a library of skills -A perception module that provides ground truth locations and visual object detection And what your left with is a glimpse into what our future will hold: robots executing tasks from human voice prompts.

AI Breakfast

33,725 görüntüleme • 3 yıl önce

Demis Hassabis confirmed every frontier AI lab is working on recursive self-improvement and in the same sentence said the safety risk of removing humans from the loop entirely keeps him up at night. That combination should stop you. The CEO of Google DeepMind just confirmed that the thing most people treat as a theoretical future risk is already the active focus of every serious lab on earth right now. He explained why it works in coding and math. The feedback loop is fast. You can verify whether an answer is correct almost instantly. You can generate synthetic training data from it. The loop closes quickly and cleanly. Then he said where it breaks down. In biology, chemistry and physics. Any domain where verifying a hypothesis requires a physical experiment in the real world. The loop does not close in seconds. It closes in weeks or months. Geoffrey Hinton said in his Nobel lecture that recursive self-improvement is the development he fears most and that once started it may not be possible to stop. Hassabis is not pushing back on that. He is describing the guardrails labs are building around a process they are already running. Every lab has to think carefully about the safety of a process where no human is in the loop. He said that as a constraint they are navigating right now. The question they are sitting with is how much of it to let run without a human watching. (Watch the full interview on YouTube at Two Minute Papers channel)
2:00

Demis Hassabis confirmed every frontier AI lab is working on recursive self-improvement and in the same sentence said the safety risk of removing humans from the loop entirely keeps him up at night. That combination should stop you. The CEO of Google DeepMind just confirmed that the thing most people treat as a theoretical future risk is already the active focus of every serious lab on earth right now. He explained why it works in coding and math. The feedback loop is fast. You can verify whether an answer is correct almost instantly. You can generate synthetic training data from it. The loop closes quickly and cleanly. Then he said where it breaks down. In biology, chemistry and physics. Any domain where verifying a hypothesis requires a physical experiment in the real world. The loop does not close in seconds. It closes in weeks or months. Geoffrey Hinton said in his Nobel lecture that recursive self-improvement is the development he fears most and that once started it may not be possible to stop. Hassabis is not pushing back on that. He is describing the guardrails labs are building around a process they are already running. Every lab has to think carefully about the safety of a process where no human is in the loop. He said that as a constraint they are navigating right now. The question they are sitting with is how much of it to let run without a human watching. (Watch the full interview on YouTube at Two Minute Papers channel)

Ihtesham Ali

68,231 görüntüleme • 1 ay önce

ChatGPT can write you a thousand trading strategies from one sentence. 999 will quietly bleed your account dry. Every single one of them looks like a winner. One method pulls the real one out. Quant desks have run it for years. It's called loop engineering. It's not a smarter prompt. It's a loop: generate a strategy, test it, score why it failed, feed that back, and run it again until one survives data it has never seen. Man Group runs $150 billion. They built an AI that writes and tests hundreds of strategies a week. And second agent whose only job is to hunt the ones that secretly overfit and kill them before a cent goes live. > describe an idea -> backtest 5 years in 12 seconds > read why it died -> feed that back in > run it again until one survives data it never saw AI can write a thousand strategies. Loop engineering is the only thing that tells you which one was ever real. 89% of retail lost money in 2025. Almost none of it from bad ideas. All of it from trading the first guess and never running it through backtest loop. Horizon just put the whole thing in a chat box, free while the beta's open: You're not losing to smarter traders. You're losing to the ones who opened the tab. Anyone can generate a strategy now. Almost no one runs the loop that proves it. That gap is the whole edge. You'll mean to check it later. Later is a "waitlist closed" page.
0:43

ChatGPT can write you a thousand trading strategies from one sentence. 999 will quietly bleed your account dry. Every single one of them looks like a winner. One method pulls the real one out. Quant desks have run it for years. It's called loop engineering. It's not a smarter prompt. It's a loop: generate a strategy, test it, score why it failed, feed that back, and run it again until one survives data it has never seen. Man Group runs $150 billion. They built an AI that writes and tests hundreds of strategies a week. And second agent whose only job is to hunt the ones that secretly overfit and kill them before a cent goes live. > describe an idea -> backtest 5 years in 12 seconds > read why it died -> feed that back in > run it again until one survives data it never saw AI can write a thousand strategies. Loop engineering is the only thing that tells you which one was ever real. 89% of retail lost money in 2025. Almost none of it from bad ideas. All of it from trading the first guess and never running it through backtest loop. Horizon just put the whole thing in a chat box, free while the beta's open: You're not losing to smarter traders. You're losing to the ones who opened the tab. Anyone can generate a strategy now. Almost no one runs the loop that proves it. That gap is the whole edge. You'll mean to check it later. Later is a "waitlist closed" page.

cvxv666

173,703 görüntüleme • 1 ay önce

Elon Musk says the robot inflection point is not a better demo. It is the moment Optimus can learn from YouTube like a human. "Humanoid robots will be the biggest product ever. The demand will be insatiable." "Who wouldn't want their own personal C-3PO, R2-D2?" Then he gives the real threshold: "If Optimus can watch videos, YouTube videos or how-to videos... and based on that video, just like a human can learn how to do that thing, then you really have task extensibility that is dramatic." "Because then it can learn anything very quickly." Right now Tesla still bootstraps with humans in mocap suits teaching primitive tasks. But Elon's endgame is self-play: "How does a child learn? A child has toys... at some point figures out how to put the triangle in the triangle hole and the circle in the circle hole." "Once you have a lot of robots, you can do this self-play... literally have the robot play with toys." That is the shift: Robots stop being programmed task-by-task. They start absorbing the internet's how-to layer, then practicing in the physical world.
2:09

Elon Musk says the robot inflection point is not a better demo. It is the moment Optimus can learn from YouTube like a human. "Humanoid robots will be the biggest product ever. The demand will be insatiable." "Who wouldn't want their own personal C-3PO, R2-D2?" Then he gives the real threshold: "If Optimus can watch videos, YouTube videos or how-to videos... and based on that video, just like a human can learn how to do that thing, then you really have task extensibility that is dramatic." "Because then it can learn anything very quickly." Right now Tesla still bootstraps with humans in mocap suits teaching primitive tasks. But Elon's endgame is self-play: "How does a child learn? A child has toys... at some point figures out how to put the triangle in the triangle hole and the circle in the circle hole." "Once you have a lot of robots, you can do this self-play... literally have the robot play with toys." That is the shift: Robots stop being programmed task-by-task. They start absorbing the internet's how-to layer, then practicing in the physical world.

Karl Mehta

27,488 görüntüleme • 1 ay önce

🚨ELON MUSK: "I thought Marxism was a dead philosophy but its alive and well in our educational institutions. Part of it is that you have a lot of colleges with these giant endowments that insulate them from reality. So you can have this fake Communist Utopia at a university because they don't have any feedback loop to reality because they have these giant endowments (and government funding too), so that allows them to be disconnected from the realities of the free market. They take advantage of the free market, but they live in a pseudo communist world. How often do we need to run the Communism experiment? It should be very obvious that the economic philosophy that is the good one is the one that does not need to build walls to keep people IN. North Korea builds walls to keep people in. East Germany had to build walls to keep people from going to West Germany. The philosophy that has to build walls to keep people in, is the bad one. And yet you see a resurgence of it, over and over again."
1:50

🚨ELON MUSK: "I thought Marxism was a dead philosophy but its alive and well in our educational institutions. Part of it is that you have a lot of colleges with these giant endowments that insulate them from reality. So you can have this fake Communist Utopia at a university because they don't have any feedback loop to reality because they have these giant endowments (and government funding too), so that allows them to be disconnected from the realities of the free market. They take advantage of the free market, but they live in a pseudo communist world. How often do we need to run the Communism experiment? It should be very obvious that the economic philosophy that is the good one is the one that does not need to build walls to keep people IN. North Korea builds walls to keep people in. East Germany had to build walls to keep people from going to West Germany. The philosophy that has to build walls to keep people in, is the bad one. And yet you see a resurgence of it, over and over again."

Autism Capital 🧩

11,140,965 görüntüleme • 1 yıl önce