Загрузка видео...

Не удалось загрузить видео

Возникла проблема при загрузке этого видео. Это может быть связано с временными проблемами сети или видео может быть недоступно.

На главную

Inside every cell, kinesin motor proteins move continuously along microtubules. Each kinesin takes 100 steps per second; at human scale, this corresponds to speeds beyond modern aircraft. Billions operate simultaneously

Science girl

5,817,742 subscribers

76,072 просмотров • 4 месяцев назад •via X (Twitter)

Наука и технологии

Anya Rossi• Live Now

Private livecam show

Комментарии: 0

Нет доступных комментариев

Здесь появятся комментарии из оригинального поста

Похожие видео

Kinesin is one of the body’s smallest transport systems: a molecular motor that walks along microscopic tracks inside your cells, carrying cargo like nutrients, enzymes, and signaling molecules to where they’re needed. 🟡 Moves energy supplies and building materials within cells 🟡 Powers nerve function by transporting molecules along axons 🟡 Keeps cell division, repair, and communication running smoothly Without kinesin, cells would lose their internal delivery network. Signals in the brain would slow, repair processes would stall, and energy wouldn’t reach where it’s needed. Kinesin turns chemistry into motion. Watching it “walk” across microtubules reveals how life maintains order and flow at the molecular level, proof that even the smallest engines inside us keep everything moving.

Kinesin is one of the body’s smallest transport systems: a molecular motor that walks along microscopic tracks inside your cells, carrying cargo like nutrients, enzymes, and signaling molecules to where they’re needed. 🟡 Moves energy supplies and building materials within cells 🟡 Powers nerve function by transporting molecules along axons 🟡 Keeps cell division, repair, and communication running smoothly Without kinesin, cells would lose their internal delivery network. Signals in the brain would slow, repair processes would stall, and energy wouldn’t reach where it’s needed. Kinesin turns chemistry into motion. Watching it “walk” across microtubules reveals how life maintains order and flow at the molecular level, proof that even the smallest engines inside us keep everything moving.

William A. Wallace, Ph.D.

70,199 просмотров • 7 месяцев назад

This is a Kinesin Molecule. These little molecular machines are commonly referred to as the "workhorse" of the cell, hauling important cargo like organelles, proteins and other cellular structures to their proper location within the cell. Kinesins are very clear & undeniable evidence of the intelligent design in Life. Kinesins are complex molecular machines, made up of 4 total proteins, each between 500-1,300 amino acids in length. If these aa sequences are not perfectly aligned from the beginning, Kinesin never forms, and cellular life would be unable to survive. Here is how they work: When a new protein, organelle, or other cellular structure is created in the Cytosol of the cell, they are constructed with built-in "binding tags," which are like shipping labels that bind to other protein molecules called Adaptor or Scaffold Proteins. These Adaptor/Scaffold Proteins then attach to the Kinesin's tail, which activates them, and then the Kinesin is guided along the microtubule track to its to its final destination. Kinesin walk on self-assembling tracks of other proteins, called microtubules, moving cargo from the inner area of the cell where they are constructed to the outer edges where they function. This is a complex & sophisticated interdependent network of molecular machines, all relying on one another to function properly for the health of the cell. This Intracellular Transportation system MUST be fully functional from the beginning - with all these working parts, or all of it fails, and the cell dies. Without this entire functioning system, Life could not exist. And the Kinesin is the centerpiece to all of it. Experiments have shown that disrupting Kinesin activity has catastrophic consequences. This type of nano-precision is an obvious example of designed engineering. Blind, unguided evolutionary processes cannot plan ahead and create complex informational highways for precision transportation. The proposed evolutionary explanation is simply "co-option." A nebulous term which basically amounts to, "We don't know how it evolved, but it must have evolved from some other thing that was similar in the past." No observational data supports evolutionary co-option. It's absurd to believe any part of this was built by blind chance. Everything in the cell points to Intelligent Design.

This is a Kinesin Molecule. These little molecular machines are commonly referred to as the "workhorse" of the cell, hauling important cargo like organelles, proteins and other cellular structures to their proper location within the cell. Kinesins are very clear & undeniable evidence of the intelligent design in Life. Kinesins are complex molecular machines, made up of 4 total proteins, each between 500-1,300 amino acids in length. If these aa sequences are not perfectly aligned from the beginning, Kinesin never forms, and cellular life would be unable to survive. Here is how they work: When a new protein, organelle, or other cellular structure is created in the Cytosol of the cell, they are constructed with built-in "binding tags," which are like shipping labels that bind to other protein molecules called Adaptor or Scaffold Proteins. These Adaptor/Scaffold Proteins then attach to the Kinesin's tail, which activates them, and then the Kinesin is guided along the microtubule track to its to its final destination. Kinesin walk on self-assembling tracks of other proteins, called microtubules, moving cargo from the inner area of the cell where they are constructed to the outer edges where they function. This is a complex & sophisticated interdependent network of molecular machines, all relying on one another to function properly for the health of the cell. This Intracellular Transportation system MUST be fully functional from the beginning - with all these working parts, or all of it fails, and the cell dies. Without this entire functioning system, Life could not exist. And the Kinesin is the centerpiece to all of it. Experiments have shown that disrupting Kinesin activity has catastrophic consequences. This type of nano-precision is an obvious example of designed engineering. Blind, unguided evolutionary processes cannot plan ahead and create complex informational highways for precision transportation. The proposed evolutionary explanation is simply "co-option." A nebulous term which basically amounts to, "We don't know how it evolved, but it must have evolved from some other thing that was similar in the past." No observational data supports evolutionary co-option. It's absurd to believe any part of this was built by blind chance. Everything in the cell points to Intelligent Design.

Divinely Designed

15,877 просмотров • 6 месяцев назад

Shane Legg, Chief AGI Scientist and co-founder of Google DeepMind, calls the human brain a "mobile processor" compared to what’s coming. - The human brain is essentially a low-power, 20-watt mobile processor constrained by biology. - While our internal neural signals crawl at just 30 meters per second, AI data travels at the speed of light. - Biological neurons typically fire at rates up to 100–200 Hz (with averages much lower, around 0.1–2 Hz, and some peaking at ~450 Hz), whereas modern silicon chips operate at clock speeds typically up to 6 billion Hz. i.e. Silicon is about 60million times faster - The physical gap is astronomical: AI hardware possesses a 6 to 8 order of magnitude advantage in speed, bandwidth, and scale. - Is human intelligence the universal upper limit? Shane Legg argues "absolutely not." - Just as humans cannot physically outlift a crane or outrun a dragster, our biological cognition cannot compete with industrial-scale compute. - As we master the architecture of intelligence, AI is mathematically destined to far exceed the capabilities of the human mind. --- From 'Google DeepMind and Hannah Fry' YT Channel (link in comment)

Shane Legg, Chief AGI Scientist and co-founder of Google DeepMind, calls the human brain a "mobile processor" compared to what’s coming. - The human brain is essentially a low-power, 20-watt mobile processor constrained by biology. - While our internal neural signals crawl at just 30 meters per second, AI data travels at the speed of light. - Biological neurons typically fire at rates up to 100–200 Hz (with averages much lower, around 0.1–2 Hz, and some peaking at ~450 Hz), whereas modern silicon chips operate at clock speeds typically up to 6 billion Hz. i.e. Silicon is about 60million times faster - The physical gap is astronomical: AI hardware possesses a 6 to 8 order of magnitude advantage in speed, bandwidth, and scale. - Is human intelligence the universal upper limit? Shane Legg argues "absolutely not." - Just as humans cannot physically outlift a crane or outrun a dragster, our biological cognition cannot compete with industrial-scale compute. - As we master the architecture of intelligence, AI is mathematically destined to far exceed the capabilities of the human mind. --- From 'Google DeepMind and Hannah Fry' YT Channel (link in comment)

Rohan Paul

236,011 просмотров • 6 месяцев назад

This is mind ATP Synthase The energy pump that powers every cell in every lifeform on Earth, slowed down by over 1,000x. Without ATP Synthase, life cannot exist. This little nano-machine is made up of a minimum of 8 distinct proteins, all designed perfectly to fit together and operate in a coordinated system that takes in ADP and turns it into ATP, which the cell uses to power metabolic processes. It's designed to only allow certain molecules in & out. Any mistakes, and it could flood the cell with toxic waste in a matter of seconds, killing everything. Each protein building block must fit perfectly together, much like all the pieces inside your phone. Without all the proteins fit & working together from the start, ATP Synthase cannot function, and Life cannot exist. But making things even more complex, ATP Synthase doesn't operate alone. It's located inside the cell membrane, where it actively works with other molecules to ensure the ATP it creates gets to the right place. It's a whole integrated network of energy movement within each cell, without which, Life could not exist. All these pieces must exist together, from the beginning, for Life to be possible. Which means it can't arise piece by piece through evolution - because the very process of evolution requires the whole system. Life requires this system in order to produce the energy it needs to replicate & evolve - but replication & evolution is supposed to have created it. You can't have one without the other. There are no simpler versions - it's literally all or nothing. This is clear evidence of an intelligently designed system. What more evidence do you need?

This is mind ATP Synthase The energy pump that powers every cell in every lifeform on Earth, slowed down by over 1,000x. Without ATP Synthase, life cannot exist. This little nano-machine is made up of a minimum of 8 distinct proteins, all designed perfectly to fit together and operate in a coordinated system that takes in ADP and turns it into ATP, which the cell uses to power metabolic processes. It's designed to only allow certain molecules in & out. Any mistakes, and it could flood the cell with toxic waste in a matter of seconds, killing everything. Each protein building block must fit perfectly together, much like all the pieces inside your phone. Without all the proteins fit & working together from the start, ATP Synthase cannot function, and Life cannot exist. But making things even more complex, ATP Synthase doesn't operate alone. It's located inside the cell membrane, where it actively works with other molecules to ensure the ATP it creates gets to the right place. It's a whole integrated network of energy movement within each cell, without which, Life could not exist. All these pieces must exist together, from the beginning, for Life to be possible. Which means it can't arise piece by piece through evolution - because the very process of evolution requires the whole system. Life requires this system in order to produce the energy it needs to replicate & evolve - but replication & evolution is supposed to have created it. You can't have one without the other. There are no simpler versions - it's literally all or nothing. This is clear evidence of an intelligently designed system. What more evidence do you need?

Divinely Designed

28,434 просмотров • 2 месяцев назад

Current robotics models really underscore how wildly efficient the human brain is. Robotics models are relatively small (low billions of parameters), are only storing a second of context, and doing 100ms inference steps. By contrast the human brain has 100 trillion synapses (roughly analogous to parameters), processes 24 frames per second, while considering hours of previous context. So how do we expect to get many order-of-magnitude improvements in robotics across all these dimensions (which are trading off against each other in terms inference compute) at the same time? And how do we explain what's going on in the brain? Is it more analogous to a software advantage or a hardware advantage? Full episode with Sergey Levine out now!

Current robotics models really underscore how wildly efficient the human brain is. Robotics models are relatively small (low billions of parameters), are only storing a second of context, and doing 100ms inference steps. By contrast the human brain has 100 trillion synapses (roughly analogous to parameters), processes 24 frames per second, while considering hours of previous context. So how do we expect to get many order-of-magnitude improvements in robotics across all these dimensions (which are trading off against each other in terms inference compute) at the same time? And how do we explain what's going on in the brain? Is it more analogous to a software advantage or a hardware advantage? Full episode with Sergey Levine out now!

Dwarkesh Patel

87,247 просмотров • 10 месяцев назад

It’s easy to overlook this… but everything in AI depends on something almost invisible. Modern transistors are measured in nanometers — where thousands fit across a single human hair. And yet, your phone contains tens of billions of them, packed into a space smaller than your fingernail. The first time you really think about it, it’s striking. Because at its core, a transistor is simple. Just a switch. On or off. 1 or 0. But when billions of these switches flip billions of times per second, they power everything: Computing. Graphics. Artificial intelligence. The contrast is fascinating. The most advanced systems we build today… rely on the simplest action, repeated at extreme scale. So here’s something I’d be curious to hear from you: As we reach atomic limits, where will the next breakthrough in computing come from? #ArtificialIntelligence #Technology #Innovation #Semiconductors #FutureOfAI

It’s easy to overlook this… but everything in AI depends on something almost invisible. Modern transistors are measured in nanometers — where thousands fit across a single human hair. And yet, your phone contains tens of billions of them, packed into a space smaller than your fingernail. The first time you really think about it, it’s striking. Because at its core, a transistor is simple. Just a switch. On or off. 1 or 0. But when billions of these switches flip billions of times per second, they power everything: Computing. Graphics. Artificial intelligence. The contrast is fascinating. The most advanced systems we build today… rely on the simplest action, repeated at extreme scale. So here’s something I’d be curious to hear from you: As we reach atomic limits, where will the next breakthrough in computing come from? #ArtificialIntelligence #Technology #Innovation #Semiconductors #FutureOfAI

Pascal Bornet

103,727 просмотров • 3 месяцев назад

This video is one of the first times I thought biology was “cool.” It shows a neutrophil cell chasing a bacterium. Originally recorded in the 1950s by David Rogers at Vanderbilt University, the video gave me a deeper appreciation for life, even at the level of a single cell, because the neutrophil's movements seem so intentful, purposeful, aware. It wasn’t until recently, though, that I actually tried to demystify the neutrophil’s movements and understand how they happen. Here's what I learned: 1. The neutrophil's surface has thousands of protein receptors. Molecules secreted by the bacteria collide with these receptors. When that happens, the proteins change shape, slightly, and initiate a signaling cascade. 2. The neutrophil “knows” where to go because of a discrepancy in bound vs. unbound receptors. The side of the cell closest to the microbe will, probabilistically, have more "bound" receptors than the other side (because the molecules secreted by the microbe have a concentration gradient). This is how the neutrophil figures out which way to move. 3. Each bound receptor activates several G proteins located inside the cell membrane. Each G protein, in turn, switches on PI3K enzymes. In this way, the original signal is amplified; a single "activated" receptor might cause ~100 copies of PI3K to get switched on downstream. 4. The PI3K enzymes stick phosphates onto lipids in the cell membrane. The side of the neutrophil facing the bacterium now has more phosphates than the "back" side. Phosphate-binding proteins, such as GEF, accumulate and then recruit Rac, thus activating it. Rac, in turn, acts like a molecular switch, ultimately recruiting Arp2/3. (TL;DR: A bunch of proteins get activated, and the high phosphate concentration at the leading edge is the key signal for all this.) 5. At any given moment, the neutrophil has millions of actin molecules. These are the proteins used to build the cytoskeleton. Half of the actins are already “assembled” into filaments, but the other half are just floating around. Arp2/3 acts as a nucleator, grabbing onto actin and then starting a new cytoskeletal branch. More actin is assembled at the leading edge (where the Arp2/3 has accumulated), where they each push on the cell membrane with ~2 piconewtons of force. Hundreds of actin chains, pushing together, causes the cell to form protrusions. 6. The assembling actin chains push the cell at a speed of ~20 micrometers per minute (the length of about ten E. coli cells placed end-to-end.) As all of this is happening, another signaling cascade, nucleated at the back end of the cell, is dismantling actin filaments and recycling them. All this happens over a span of about 30 seconds. Much of this process is invisible; what we see, instead, is "just" a cell chasing its prey. But that's the wonderful thing about biology: A singular observation is usually more than enough fodder for a lifetime of work. The well is deep. There is always more to learn.

This video is one of the first times I thought biology was “cool.” It shows a neutrophil cell chasing a bacterium. Originally recorded in the 1950s by David Rogers at Vanderbilt University, the video gave me a deeper appreciation for life, even at the level of a single cell, because the neutrophil's movements seem so intentful, purposeful, aware. It wasn’t until recently, though, that I actually tried to demystify the neutrophil’s movements and understand how they happen. Here's what I learned: 1. The neutrophil's surface has thousands of protein receptors. Molecules secreted by the bacteria collide with these receptors. When that happens, the proteins change shape, slightly, and initiate a signaling cascade. 2. The neutrophil “knows” where to go because of a discrepancy in bound vs. unbound receptors. The side of the cell closest to the microbe will, probabilistically, have more "bound" receptors than the other side (because the molecules secreted by the microbe have a concentration gradient). This is how the neutrophil figures out which way to move. 3. Each bound receptor activates several G proteins located inside the cell membrane. Each G protein, in turn, switches on PI3K enzymes. In this way, the original signal is amplified; a single "activated" receptor might cause ~100 copies of PI3K to get switched on downstream. 4. The PI3K enzymes stick phosphates onto lipids in the cell membrane. The side of the neutrophil facing the bacterium now has more phosphates than the "back" side. Phosphate-binding proteins, such as GEF, accumulate and then recruit Rac, thus activating it. Rac, in turn, acts like a molecular switch, ultimately recruiting Arp2/3. (TL;DR: A bunch of proteins get activated, and the high phosphate concentration at the leading edge is the key signal for all this.) 5. At any given moment, the neutrophil has millions of actin molecules. These are the proteins used to build the cytoskeleton. Half of the actins are already “assembled” into filaments, but the other half are just floating around. Arp2/3 acts as a nucleator, grabbing onto actin and then starting a new cytoskeletal branch. More actin is assembled at the leading edge (where the Arp2/3 has accumulated), where they each push on the cell membrane with ~2 piconewtons of force. Hundreds of actin chains, pushing together, causes the cell to form protrusions. 6. The assembling actin chains push the cell at a speed of ~20 micrometers per minute (the length of about ten E. coli cells placed end-to-end.) As all of this is happening, another signaling cascade, nucleated at the back end of the cell, is dismantling actin filaments and recycling them. All this happens over a span of about 30 seconds. Much of this process is invisible; what we see, instead, is "just" a cell chasing its prey. But that's the wonderful thing about biology: A singular observation is usually more than enough fodder for a lifetime of work. The well is deep. There is always more to learn.

Niko McCarty.

127,178 просмотров • 6 месяцев назад

Changpeng Zhao just identified the next financial market, and every traditional bank is structurally obsolete for it. “Transacting millions of times more than humans.” The Machine Economy. Financial systems were built for human speeds. We buy coffee, pay rent, sleep. AI agents operate continuously without rest. Zhao: “You’ll soon have thousands of AI agents working for you.” Negotiating APIs, buying compute, booking logistics, executing trades. All simultaneously at machine speed. Transaction volume from AI commerce won’t supplement human activity. It’ll eclipse it completely. The infrastructure isn’t Visa or banks. It’s stablecoins like USDC. AI needs programmable money settling instantly, not in “3-5 business days.” Zhao: “Digital assistants will not find banks suitable.” AI executing millisecond arbitrage can’t wait for bank transfers. Delays don’t slow the system. They break it entirely. From high-frequency trading to high-frequency living. Banking adapts or gets routed around. Zhao: “They’ll book hotels, buy tickets… transacting millions of times.” Credit cards need identity. Stablecoins need cryptographic keys. AI excels at key management, not waiting for human approval processes. Default currency for AI won’t be physical dollars. It’ll be programmable dollars. First institution integrating with AI agents captures next-century transaction flow. Legacy banking assumes human speeds with settlement delays. AI operates at microsecond velocity with instant settlement requirements. Architecturally incompatible. Infrastructure handling trillions in human commerce becomes the bottleneck strangling quadrillions in machine commerce. Markets eliminate bottlenecks by building parallel systems. Once machine economy reaches scale, traditional banking doesn’t compete. It becomes the slow legacy option only humans use while everything significant happens in infrastructure they can’t access at competitive speeds.

Changpeng Zhao just identified the next financial market, and every traditional bank is structurally obsolete for it. “Transacting millions of times more than humans.” The Machine Economy. Financial systems were built for human speeds. We buy coffee, pay rent, sleep. AI agents operate continuously without rest. Zhao: “You’ll soon have thousands of AI agents working for you.” Negotiating APIs, buying compute, booking logistics, executing trades. All simultaneously at machine speed. Transaction volume from AI commerce won’t supplement human activity. It’ll eclipse it completely. The infrastructure isn’t Visa or banks. It’s stablecoins like USDC. AI needs programmable money settling instantly, not in “3-5 business days.” Zhao: “Digital assistants will not find banks suitable.” AI executing millisecond arbitrage can’t wait for bank transfers. Delays don’t slow the system. They break it entirely. From high-frequency trading to high-frequency living. Banking adapts or gets routed around. Zhao: “They’ll book hotels, buy tickets… transacting millions of times.” Credit cards need identity. Stablecoins need cryptographic keys. AI excels at key management, not waiting for human approval processes. Default currency for AI won’t be physical dollars. It’ll be programmable dollars. First institution integrating with AI agents captures next-century transaction flow. Legacy banking assumes human speeds with settlement delays. AI operates at microsecond velocity with instant settlement requirements. Architecturally incompatible. Infrastructure handling trillions in human commerce becomes the bottleneck strangling quadrillions in machine commerce. Markets eliminate bottlenecks by building parallel systems. Once machine economy reaches scale, traditional banking doesn’t compete. It becomes the slow legacy option only humans use while everything significant happens in infrastructure they can’t access at competitive speeds.

Dustin

29,893 просмотров • 5 месяцев назад

Before dawn over Armenia, the sky starts to look almost alive. This timelapse was captured westbound as we crossed one of the busiest, and most geopolitically sensitive pieces of airspace on Earth. To the south lies Iran. To the north, Russia. Between them, a narrow corridor through the Caucasus carrying hundreds of aircraft linking Europe, the Gulf, and Asia. Every light you see here is another aircraft, another crew, another story moving through the night. What makes it remarkable is the precision. At closing speeds approaching 1,000 mph, aircraft pass each other separated not by miles, but by carefully managed vertical layers of airspace. Invisible highways in the sky, built on GPS accuracy, disciplined procedures, satellite surveillance, and trust. And yet these routes are never static. Politics shapes them. Conflict bends them. Closed airspace redraws them almost overnight. What was once a straightforward great-circle route can quickly become a carefully threaded path between terrain, and diplomacy. From the cockpit, though, there’s a strange beauty to it all. Tiny lights crossing ancient landscapes. Modern aircraft moving through corridors shaped by geography and history alike. Controlled. Precise. Fleeting. A reminder that aviation is never just about the aircraft. It’s about learning how to move safely through an increasingly complicated world.

Before dawn over Armenia, the sky starts to look almost alive. This timelapse was captured westbound as we crossed one of the busiest, and most geopolitically sensitive pieces of airspace on Earth. To the south lies Iran. To the north, Russia. Between them, a narrow corridor through the Caucasus carrying hundreds of aircraft linking Europe, the Gulf, and Asia. Every light you see here is another aircraft, another crew, another story moving through the night. What makes it remarkable is the precision. At closing speeds approaching 1,000 mph, aircraft pass each other separated not by miles, but by carefully managed vertical layers of airspace. Invisible highways in the sky, built on GPS accuracy, disciplined procedures, satellite surveillance, and trust. And yet these routes are never static. Politics shapes them. Conflict bends them. Closed airspace redraws them almost overnight. What was once a straightforward great-circle route can quickly become a carefully threaded path between terrain, and diplomacy. From the cockpit, though, there’s a strange beauty to it all. Tiny lights crossing ancient landscapes. Modern aircraft moving through corridors shaped by geography and history alike. Controlled. Precise. Fleeting. A reminder that aviation is never just about the aircraft. It’s about learning how to move safely through an increasingly complicated world.

Scott Bateman MBE

43,390 просмотров • 2 месяцев назад

A new Nature paper from Johns Hopkins (by Prof. Lin Dingchang Lin ) just solved one of the hardest problems in biology: how do you record what every cell in a tissue experienced over time, not just what it looks like right now? The answer: GEMINI — Granularly Expanding Memory for Intracellular Narrative Integration. It works exactly like tree rings. Cells are genetically engineered to express a computationally designed protein assembly. As the assembly grows inside the cell, it captures cellular activity as fluorescent ring patterns — each ring a timestamp, each ring's properties encoding signal intensity. Look at a cross-section under a microscope and you can read the cell's history backward, with ~15-minute resolution. The key: cells build the recorder themselves. GEMINI doesn't interfere with normal function — it just quietly writes. What they demonstrated: In a full tumor xenograft, GEMINI captured every cancer cell's activity history across the entire tumor while it continued to grow normally. For the first time, researchers can look back and see how different regions of the same tumor responded differently to therapy over time — not snapshots, but film. In a mouse brain, GEMINI recorded neural activity dynamics without disrupting behavior, coordination, or memory. It could temporally resolve the history of a brain seizure. Why this matters: Every tool we have in biology gives you state — what the cell looks like now. Sequencing, imaging, proteomics — all snapshots. GEMINI gives you trajectory. It's the difference between a photograph and a video, applied to every cell in an organ simultaneously. The team is explicit that AI-based decoding tools will be central to reading GEMINI's output at whole-brain scale. This is the data layer that makes temporal single-cell atlases possible. Paper: Congratulations Dingchang Lin

A new Nature paper from Johns Hopkins (by Prof. Lin Dingchang Lin ) just solved one of the hardest problems in biology: how do you record what every cell in a tissue experienced over time, not just what it looks like right now? The answer: GEMINI — Granularly Expanding Memory for Intracellular Narrative Integration. It works exactly like tree rings. Cells are genetically engineered to express a computationally designed protein assembly. As the assembly grows inside the cell, it captures cellular activity as fluorescent ring patterns — each ring a timestamp, each ring's properties encoding signal intensity. Look at a cross-section under a microscope and you can read the cell's history backward, with ~15-minute resolution. The key: cells build the recorder themselves. GEMINI doesn't interfere with normal function — it just quietly writes. What they demonstrated: In a full tumor xenograft, GEMINI captured every cancer cell's activity history across the entire tumor while it continued to grow normally. For the first time, researchers can look back and see how different regions of the same tumor responded differently to therapy over time — not snapshots, but film. In a mouse brain, GEMINI recorded neural activity dynamics without disrupting behavior, coordination, or memory. It could temporally resolve the history of a brain seizure. Why this matters: Every tool we have in biology gives you state — what the cell looks like now. Sequencing, imaging, proteomics — all snapshots. GEMINI gives you trajectory. It's the difference between a photograph and a video, applied to every cell in an organ simultaneously. The team is explicit that AI-based decoding tools will be central to reading GEMINI's output at whole-brain scale. This is the data layer that makes temporal single-cell atlases possible. Paper: Congratulations Dingchang Lin

Bo Wang

85,095 просмотров • 4 месяцев назад

Elon Musk reveals Tesla is building a 30,000-robot academy where humanoids learn from each other. Cars were easy. Tesla had ten million on the road, beaming back driving data every second. But humanoid robots? There weren't ten million Optimi yet. There weren't ten. Robotics had run data-starved for decades. Tesla decided to fix it. You couldn't train a humanoid that had never been deployed. So Musk built a school for them instead. "We can have at least 10,000 Optimus robots, maybe 20-30,000, that are doing self-play and testing different tasks." Tesla called it the Optimus Academy. Picture a warehouse the size of a chip fab. Thirty thousand humanoid robots inside. Picking things up. Folding clothes. Walking. Tripping. Catching themselves. Failing in ways no human roboticist had thought to script. Each watching the others, learning what the human body shouldn't have made look easy. Every move generated a data point. Every failure generated a sample. Every robot taught every other robot. In simulation, Tesla could spin up a million robots overnight. But simulated physics lied about friction, slip, and drift. Real physics didn't. Cars learned from drivers. Optimi learned from each other. Each generation made the next one cheaper, faster, smarter. By the tenth generation, no human would recognize the curriculum. Recursive learning at electromechanical scale. Musk, on closing the loop: "You use the tens of thousands of robots in the real world to close the simulation to reality gap." Whoever opened the academy first owned the species. P.S. I made a playbook breaking down 100+ most powerful decision making mental models used by history's greatest thinkers. 5,000+ downloads. 113 five-star reviews. Grab a free copy here: If you're new here, follow GeniusThinking for content on the greatest minds in economics, psychology, and history. — Elon Musk ( Elon Musk ), CEO of Tesla and SpaceX, on Dwarkesh Patel's ( Dwarkesh Patel ) podcast

Elon Musk reveals Tesla is building a 30,000-robot academy where humanoids learn from each other. Cars were easy. Tesla had ten million on the road, beaming back driving data every second. But humanoid robots? There weren't ten million Optimi yet. There weren't ten. Robotics had run data-starved for decades. Tesla decided to fix it. You couldn't train a humanoid that had never been deployed. So Musk built a school for them instead. "We can have at least 10,000 Optimus robots, maybe 20-30,000, that are doing self-play and testing different tasks." Tesla called it the Optimus Academy. Picture a warehouse the size of a chip fab. Thirty thousand humanoid robots inside. Picking things up. Folding clothes. Walking. Tripping. Catching themselves. Failing in ways no human roboticist had thought to script. Each watching the others, learning what the human body shouldn't have made look easy. Every move generated a data point. Every failure generated a sample. Every robot taught every other robot. In simulation, Tesla could spin up a million robots overnight. But simulated physics lied about friction, slip, and drift. Real physics didn't. Cars learned from drivers. Optimi learned from each other. Each generation made the next one cheaper, faster, smarter. By the tenth generation, no human would recognize the curriculum. Recursive learning at electromechanical scale. Musk, on closing the loop: "You use the tens of thousands of robots in the real world to close the simulation to reality gap." Whoever opened the academy first owned the species. P.S. I made a playbook breaking down 100+ most powerful decision making mental models used by history's greatest thinkers. 5,000+ downloads. 113 five-star reviews. Grab a free copy here: If you're new here, follow GeniusThinking for content on the greatest minds in economics, psychology, and history. — Elon Musk ( Elon Musk ), CEO of Tesla and SpaceX, on Dwarkesh Patel's ( Dwarkesh Patel ) podcast

GeniusThinking

136,378 просмотров • 1 месяц назад

The bacterial flagellum looks like a simple tail, or whip. But it’s actually a rotating motor, and perhaps the most sophisticated protein complex nature has ever evolved. In e. coli, these motors are capable of astonishing speeds; about 15,000 rpm. (The world record, according to one study, is for a Vibrio cell that was “clocked at 100,000 rpm by laser microscopy.) The flagellum propels the cell forward at speeds of 20-30 microns per second, or roughly 15 body lengths per second. If scaled up to the size of a cheetah, E. coli would *nearly* be the fastest land organism. The darting movements of a microbe were first observed in 1676 by Antony van Leeuwenhoek, a Dutch cloth merchant. Antony was delighted by the motion of his “animalcules,” writing: “I must say, for my part, that no more pleasant sight has ever yet come before my eye than these many thousands of living creatures, seen all alive in a little drop of water, moving among one another, each several creature having its own proper motion.” But Leeuwenhoek did not see flagella. He assumed, rather, that these animalcules must be “furnished with paws” instead. Christian Ehrenberg would not properly describe flagella until 1836. But amazingly, all the way up until the 1970s, nobody actually knew how the flagellum spun! In 1973, there were two competing models people argued over: the helical-wave (bending) model and the rotating (corkscrew) model. The first model suggested that the flagellum whipped back and forth, side-to-side, to propel the cell like paddle. The corkscrew model suggested that the whole flagellum instead spins around like a screw. In 1974, the corkscrew model finally won out. For two separate studies, scientists affixed flagella to glass slides using antibodies, and watched as the cells spun around and around like corkscrews. And finally, in just the last year, high-resolution structures of the flagellum have revealed a LOT more about its intricate assembly. The tail is made from ~20,000 self-assembling copies of a single protein, called flagellin. A “driveshaft,” or rod, spins the tail and is itself made of 26 protein subunits. Each “motor” in E. coli consists of 11 stators, each of which is made from 7 proteins.(Other types of cells have even more stators, and swim with much higher torques.) The flagellum spins when protons flow into the cell through tiny channels in these stators; akin to water running through a turbine. Each proton makes a small part of the stator change shape and push against the rotor, nudging it forward one step. With dozens of stators working at once, these nudges quickly spin the propeller. I'm writing an essay for Asimov Press about this now, and am really enjoying learning about the flagellum and its history. It's an extraordinarily complicated structure, though, and has been a challenge to understand!

The bacterial flagellum looks like a simple tail, or whip. But it’s actually a rotating motor, and perhaps the most sophisticated protein complex nature has ever evolved. In e. coli, these motors are capable of astonishing speeds; about 15,000 rpm. (The world record, according to one study, is for a Vibrio cell that was “clocked at 100,000 rpm by laser microscopy.) The flagellum propels the cell forward at speeds of 20-30 microns per second, or roughly 15 body lengths per second. If scaled up to the size of a cheetah, E. coli would nearly be the fastest land organism. The darting movements of a microbe were first observed in 1676 by Antony van Leeuwenhoek, a Dutch cloth merchant. Antony was delighted by the motion of his “animalcules,” writing: “I must say, for my part, that no more pleasant sight has ever yet come before my eye than these many thousands of living creatures, seen all alive in a little drop of water, moving among one another, each several creature having its own proper motion.” But Leeuwenhoek did not see flagella. He assumed, rather, that these animalcules must be “furnished with paws” instead. Christian Ehrenberg would not properly describe flagella until 1836. But amazingly, all the way up until the 1970s, nobody actually knew how the flagellum spun! In 1973, there were two competing models people argued over: the helical-wave (bending) model and the rotating (corkscrew) model. The first model suggested that the flagellum whipped back and forth, side-to-side, to propel the cell like paddle. The corkscrew model suggested that the whole flagellum instead spins around like a screw. In 1974, the corkscrew model finally won out. For two separate studies, scientists affixed flagella to glass slides using antibodies, and watched as the cells spun around and around like corkscrews. And finally, in just the last year, high-resolution structures of the flagellum have revealed a LOT more about its intricate assembly. The tail is made from ~20,000 self-assembling copies of a single protein, called flagellin. A “driveshaft,” or rod, spins the tail and is itself made of 26 protein subunits. Each “motor” in E. coli consists of 11 stators, each of which is made from 7 proteins.(Other types of cells have even more stators, and swim with much higher torques.) The flagellum spins when protons flow into the cell through tiny channels in these stators; akin to water running through a turbine. Each proton makes a small part of the stator change shape and push against the rotor, nudging it forward one step. With dozens of stators working at once, these nudges quickly spin the propeller. I'm writing an essay for Asimov Press about this now, and am really enjoying learning about the flagellum and its history. It's an extraordinarily complicated structure, though, and has been a challenge to understand!

Niko McCarty.

51,893 просмотров • 10 месяцев назад

A rare insight into the AN/PRC-10 field radio used in Korea and early Vietnam war, mainly built by General Electric. A 1950s vacuum tube man-pack system, fully assembled and wired by hand, built around 16 subminiature vacuum tubes with no semiconductors, no digital control, and no onboard computation. To operate, it required a stacked multi voltage battery system producing 1.5V, 6V, 67.5V and 135V simultaneously, just to keep analog amplification and RF stages stable in field conditions. For scale, a modern smartphone battery like an iPhone stores roughly 10 to 12 watt hours of energy in a compact lithium cell. The PRC-10 was operating with far lower usable power output, around 1 watt of transmission power, but in a completely discrete high voltage analog architecture. Each unit costed roughly $300 to $430 in 1950s terms, about $4K to $5K today, not because of electronics complexity in the modern sense, but because every radio had to be individually hand calibrated for frequency stability and field reliability all built by hand. In practice, systems like this gave US infantry units reliable 5 to 12 km communication in terrain where coordination with artillery and manoeuvre units directly influenced survival outcomes. What looks like simple battlefield equipment today was a fully hand built analog communication system that carried command decisions across entire units in real time. 🎥Biofab138

A rare insight into the AN/PRC-10 field radio used in Korea and early Vietnam war, mainly built by General Electric. A 1950s vacuum tube man-pack system, fully assembled and wired by hand, built around 16 subminiature vacuum tubes with no semiconductors, no digital control, and no onboard computation. To operate, it required a stacked multi voltage battery system producing 1.5V, 6V, 67.5V and 135V simultaneously, just to keep analog amplification and RF stages stable in field conditions. For scale, a modern smartphone battery like an iPhone stores roughly 10 to 12 watt hours of energy in a compact lithium cell. The PRC-10 was operating with far lower usable power output, around 1 watt of transmission power, but in a completely discrete high voltage analog architecture. Each unit costed roughly $300 to $430 in 1950s terms, about $4K to $5K today, not because of electronics complexity in the modern sense, but because every radio had to be individually hand calibrated for frequency stability and field reliability all built by hand. In practice, systems like this gave US infantry units reliable 5 to 12 km communication in terrain where coordination with artillery and manoeuvre units directly influenced survival outcomes. What looks like simple battlefield equipment today was a fully hand built analog communication system that carried command decisions across entire units in real time. 🎥Biofab138

Ammanichanda

37,972 просмотров • 15 дней назад

Elon Musk just made the boldest product prediction in modern business: Optimus will be 10 times larger than the biggest product ever created. Not comparable to the largest products. Ten times bigger than whatever holds that record now. Musk: “I think it will be 10 times bigger than the next biggest product ever made.” The claim sounds delusional until you trace the logic. Mass-market humanoid robots need three things simultaneously: real intelligence, natural language understanding, and manufacturing at scale with mass-market pricing. That combination is almost impossible. Most companies have one element, maybe two. Nobody else has all three solved. Musk: “Tesla is the only company with all the required ingredients.” AI trained on physical world tasks, not just text. Vision systems proven across millions of vehicles. Autonomous decision-making deployed at scale. Manufacturing capability producing complex products in millions of units affordably. Cost discipline through vertical integration. Every required component already exists inside Tesla, developed for cars but transferable to humanoid form. Musk: “You should be able to ask them to do things naturally.” When robots understand natural commands and execute tasks autonomously, usefulness becomes universal. Not specialized equipment for factories. General-purpose capability adapting to any physical task you need done. Universal usefulness means market size stops being defined by applications. It becomes defined by human labor itself. Musk: “Nothing will even be close.” The prediction isn’t that Optimus succeeds. It’s that success creates commercial value exceeding everything in history by an order of magnitude. Useful robots at consumer prices don’t sell in millions. They sell in billions. Every home. Every business. Every job currently requiring human physical presence. The addressable market becomes all human labor. And human labor is the largest market in existence. Tesla possesses the complete capability stack. AI, manufacturing, cost optimization, real-world deployment experience. The exact combination required to build genuinely useful robots at prices enabling mass adoption. If the prediction holds, this isn’t a product category. It’s the product category that makes everything else look like rounding errors. The iPhone revolutionized communication and sold over 2 billion units. Optimus targets physical labor, a market orders of magnitude larger, with comparable adoption potential if the capability delivers. Ten times bigger than the next biggest product isn’t hyperbole if you’re solving human physical labor at consumer economics. It’s just math on what happens when scarcity in the largest market disappears.

Elon Musk just made the boldest product prediction in modern business: Optimus will be 10 times larger than the biggest product ever created. Not comparable to the largest products. Ten times bigger than whatever holds that record now. Musk: “I think it will be 10 times bigger than the next biggest product ever made.” The claim sounds delusional until you trace the logic. Mass-market humanoid robots need three things simultaneously: real intelligence, natural language understanding, and manufacturing at scale with mass-market pricing. That combination is almost impossible. Most companies have one element, maybe two. Nobody else has all three solved. Musk: “Tesla is the only company with all the required ingredients.” AI trained on physical world tasks, not just text. Vision systems proven across millions of vehicles. Autonomous decision-making deployed at scale. Manufacturing capability producing complex products in millions of units affordably. Cost discipline through vertical integration. Every required component already exists inside Tesla, developed for cars but transferable to humanoid form. Musk: “You should be able to ask them to do things naturally.” When robots understand natural commands and execute tasks autonomously, usefulness becomes universal. Not specialized equipment for factories. General-purpose capability adapting to any physical task you need done. Universal usefulness means market size stops being defined by applications. It becomes defined by human labor itself. Musk: “Nothing will even be close.” The prediction isn’t that Optimus succeeds. It’s that success creates commercial value exceeding everything in history by an order of magnitude. Useful robots at consumer prices don’t sell in millions. They sell in billions. Every home. Every business. Every job currently requiring human physical presence. The addressable market becomes all human labor. And human labor is the largest market in existence. Tesla possesses the complete capability stack. AI, manufacturing, cost optimization, real-world deployment experience. The exact combination required to build genuinely useful robots at prices enabling mass adoption. If the prediction holds, this isn’t a product category. It’s the product category that makes everything else look like rounding errors. The iPhone revolutionized communication and sold over 2 billion units. Optimus targets physical labor, a market orders of magnitude larger, with comparable adoption potential if the capability delivers. Ten times bigger than the next biggest product isn’t hyperbole if you’re solving human physical labor at consumer economics. It’s just math on what happens when scarcity in the largest market disappears.

Dustin

26,722 просмотров • 5 месяцев назад

🚨🇺🇸🇮🇷 HOW HARD IS IT FOR IRAN TO HIT A U.S AIRCRAFT CARRIER? Critics love to call aircraft carriers sitting ducks. The argument sounds simple enough: a ship the size of a small city must be easy to find, easy to hit, and devastating to lose. In the age of hypersonic missiles and satellite surveillance, some analysts insist the carrier is already obsolete, but reality is far less dramatic. Aircraft carriers are not lonely targets drifting across the ocean. They operate inside one of the most sophisticated defensive systems ever built, something the U.S. Navy calls “defense in depth.” Think of it less like a single ship and more like a moving fortress. The first layer is the carrier strike group itself. Destroyers and cruisers surround the carrier, each equipped with the Aegis combat system, a network of radar and missiles capable of tracking hundreds of threats simultaneously. Incoming missiles can be intercepted hundreds of miles away, often long before the carrier itself is even in danger. The second layer lives in the sky. Aircraft like the E-2D Hawkeye patrol high above the fleet, acting as airborne radar stations that can spot low-flying missiles or enemy aircraft long before ship-based sensors could. If something suspicious appears, fighter jets such as F/A-18s or F-35s can intercept the threat before a shot is even fired. In other words, the battle is pushed far away from the carrier itself. If a missile somehow slips through those outer layers, electronic warfare becomes the next shield. Modern carriers can jam or confuse a missile’s guidance system, essentially blinding it or feeding it false information. Chaff clouds and flares create fake targets in the sky, turning the missile’s final seconds into a guessing game. Sometimes the missile never finds the ship at all. And then there is the last line of defense. Close-range interceptors like Sea Sparrow and Rolling Airframe Missiles can shoot down threats at the final moment. If everything else fails, the Phalanx close-in weapon system, a rapid-fire Gatling gun often described as a “wall of lead,” can tear an incoming missile apart just seconds before impact. None of this means aircraft carriers are invincible. Every military system has vulnerabilities, and modern anti-ship weapons are increasingly capable. But the idea that carriers are easy targets, helpless giants waiting to be sunk, misunderstands how they actually operate. A carrier is not just a ship. It's a layered defense network, a mobile airbase, and the center of an entire fleet designed to make hitting it one of the most difficult tasks in modern warfare.

🚨🇺🇸🇮🇷 HOW HARD IS IT FOR IRAN TO HIT A U.S AIRCRAFT CARRIER? Critics love to call aircraft carriers sitting ducks. The argument sounds simple enough: a ship the size of a small city must be easy to find, easy to hit, and devastating to lose. In the age of hypersonic missiles and satellite surveillance, some analysts insist the carrier is already obsolete, but reality is far less dramatic. Aircraft carriers are not lonely targets drifting across the ocean. They operate inside one of the most sophisticated defensive systems ever built, something the U.S. Navy calls “defense in depth.” Think of it less like a single ship and more like a moving fortress. The first layer is the carrier strike group itself. Destroyers and cruisers surround the carrier, each equipped with the Aegis combat system, a network of radar and missiles capable of tracking hundreds of threats simultaneously. Incoming missiles can be intercepted hundreds of miles away, often long before the carrier itself is even in danger. The second layer lives in the sky. Aircraft like the E-2D Hawkeye patrol high above the fleet, acting as airborne radar stations that can spot low-flying missiles or enemy aircraft long before ship-based sensors could. If something suspicious appears, fighter jets such as F/A-18s or F-35s can intercept the threat before a shot is even fired. In other words, the battle is pushed far away from the carrier itself. If a missile somehow slips through those outer layers, electronic warfare becomes the next shield. Modern carriers can jam or confuse a missile’s guidance system, essentially blinding it or feeding it false information. Chaff clouds and flares create fake targets in the sky, turning the missile’s final seconds into a guessing game. Sometimes the missile never finds the ship at all. And then there is the last line of defense. Close-range interceptors like Sea Sparrow and Rolling Airframe Missiles can shoot down threats at the final moment. If everything else fails, the Phalanx close-in weapon system, a rapid-fire Gatling gun often described as a “wall of lead,” can tear an incoming missile apart just seconds before impact. None of this means aircraft carriers are invincible. Every military system has vulnerabilities, and modern anti-ship weapons are increasingly capable. But the idea that carriers are easy targets, helpless giants waiting to be sunk, misunderstands how they actually operate. A carrier is not just a ship. It's a layered defense network, a mobile airbase, and the center of an entire fleet designed to make hitting it one of the most difficult tasks in modern warfare.

Mario Nawfal

1,159,785 просмотров • 4 месяцев назад

🇹🇷 Ottoman Empire at Sea? Turkey’s 50 Warship + Carrier Mega Build Signals a New Mediterranean Superpower Era Turkey is in the middle of one of the most aggressive naval expansions in modern military history, with around 50 warships under construction simultaneously across multiple shipyards. But this isn’t just a fleet expansion, it’s a full spectrum maritime transformation. From stealth capable frigates and corvettes to amphibious assault ships, submarines, support vessels, and a future aircraft carrier program, Ankara is rapidly pushing its navy toward true blue water capability, designed to operate far beyond its own coastline. Defense analysts say the scale and speed of this buildup is not about defense alone, but about long term regional power projection across the Eastern Mediterranean, Black Sea, and wider strategic waterways. The comparison with Israel is increasingly unavoidable. Israel maintains one of the most advanced and combat proven air forces in the Middle East, giving it a dominant edge in airpower and precision strike capability. However, Turkey’s strategy appears focused on balancing that advantage through overwhelming maritime expansion, naval aviation development, and indigenous defense production. At the same time, Ankara is accelerating development of its own next generation fighter programs and expanding domestic aerospace capacity, signaling an ambition to control not just the seas, but also the skies over future conflict zones. The debate inside strategic circles is intensifying. Supporters call it sovereign military independence. Critics see a long term geopolitical project aimed at restoring influence across former Ottoman spheres of power. With aircraft carrier ambitions, mass shipbuilding output, and a rapidly expanding defense industry, Turkey is no longer just upgrading its navy, it is rewriting its strategic identity. The question now is no longer whether Turkey is rising… but how far it intends to go.

🇹🇷 Ottoman Empire at Sea? Turkey’s 50 Warship + Carrier Mega Build Signals a New Mediterranean Superpower Era Turkey is in the middle of one of the most aggressive naval expansions in modern military history, with around 50 warships under construction simultaneously across multiple shipyards. But this isn’t just a fleet expansion, it’s a full spectrum maritime transformation. From stealth capable frigates and corvettes to amphibious assault ships, submarines, support vessels, and a future aircraft carrier program, Ankara is rapidly pushing its navy toward true blue water capability, designed to operate far beyond its own coastline. Defense analysts say the scale and speed of this buildup is not about defense alone, but about long term regional power projection across the Eastern Mediterranean, Black Sea, and wider strategic waterways. The comparison with Israel is increasingly unavoidable. Israel maintains one of the most advanced and combat proven air forces in the Middle East, giving it a dominant edge in airpower and precision strike capability. However, Turkey’s strategy appears focused on balancing that advantage through overwhelming maritime expansion, naval aviation development, and indigenous defense production. At the same time, Ankara is accelerating development of its own next generation fighter programs and expanding domestic aerospace capacity, signaling an ambition to control not just the seas, but also the skies over future conflict zones. The debate inside strategic circles is intensifying. Supporters call it sovereign military independence. Critics see a long term geopolitical project aimed at restoring influence across former Ottoman spheres of power. With aircraft carrier ambitions, mass shipbuilding output, and a rapidly expanding defense industry, Turkey is no longer just upgrading its navy, it is rewriting its strategic identity. The question now is no longer whether Turkey is rising… but how far it intends to go.

War Radar

32,524 просмотров • 27 дней назад

Strategic Bombers Are More Relevant Than Ever Decades ago, these aircraft had to fly directly over their targets to drop bombs. That era is over, and things have only gotten easier for them. Now they are standoff platforms capable of launching modern, heavy, long-range missiles (thousands of kilometers), including hypersonics, in large quantities, without ever entering contested airspace. This proves they are far from obsolete and have become massive flying weapons platforms. These bombers operate in squadrons, with missile ranges of 1,000–5,500 km (cruise) or around 2,000 km (hypersonic), devastating warheads, and launched in high numbers. Hypersonics like Kinzhal, YJ-21, and the future ARRW can evade modern defenses, while cruise missiles like JASSM and Kh-101 achieve saturation at a much lower cost. Russian glide bombs are being motorized to become an even cheaper option, and there are already configurations where bombers carry up to 20 glide bombs. Imagine 5 B-52H Stratofortress launching 20–24 stealth AGM-158 JASSM-ER missiles with 1,000 km range each. That’s over 100 missiles in a single salvo. Or 5 Tu-160M2 launching 6 Kh-47M2 Kinzhal each, or in another configuration up to 12 Kh-101 missiles per aircraft. If you want to go further: what navy could survive a squadron salvo from Chinese H-6 or H-20 launching 4–6 anti-ship hypersonic missiles with 1,500–2,000 km range? Strategic bombers are more relevant than ever and now operate much more safely. In the future, in possible larger conflicts, we’ll see waves of 20 bombers launching salvos of 150–250 missiles, overwhelming defenses and causing immense destruction. These platforms provide extended reach at a far lower cost than medium- or long-range ground launched missiles. Not to mention the low operating costs of the B-52, Tu-95, and H-6, with the latter probably having even lower costs. The trend is toward missiles and bombs with ever-greater range, lower cost, and modular designs to be launched from bombers. We can expect these aircraft to remain in service for at least another 20–30 years.

Strategic Bombers Are More Relevant Than Ever Decades ago, these aircraft had to fly directly over their targets to drop bombs. That era is over, and things have only gotten easier for them. Now they are standoff platforms capable of launching modern, heavy, long-range missiles (thousands of kilometers), including hypersonics, in large quantities, without ever entering contested airspace. This proves they are far from obsolete and have become massive flying weapons platforms. These bombers operate in squadrons, with missile ranges of 1,000–5,500 km (cruise) or around 2,000 km (hypersonic), devastating warheads, and launched in high numbers. Hypersonics like Kinzhal, YJ-21, and the future ARRW can evade modern defenses, while cruise missiles like JASSM and Kh-101 achieve saturation at a much lower cost. Russian glide bombs are being motorized to become an even cheaper option, and there are already configurations where bombers carry up to 20 glide bombs. Imagine 5 B-52H Stratofortress launching 20–24 stealth AGM-158 JASSM-ER missiles with 1,000 km range each. That’s over 100 missiles in a single salvo. Or 5 Tu-160M2 launching 6 Kh-47M2 Kinzhal each, or in another configuration up to 12 Kh-101 missiles per aircraft. If you want to go further: what navy could survive a squadron salvo from Chinese H-6 or H-20 launching 4–6 anti-ship hypersonic missiles with 1,500–2,000 km range? Strategic bombers are more relevant than ever and now operate much more safely. In the future, in possible larger conflicts, we’ll see waves of 20 bombers launching salvos of 150–250 missiles, overwhelming defenses and causing immense destruction. These platforms provide extended reach at a far lower cost than medium- or long-range ground launched missiles. Not to mention the low operating costs of the B-52, Tu-95, and H-6, with the latter probably having even lower costs. The trend is toward missiles and bombs with ever-greater range, lower cost, and modular designs to be launched from bombers. We can expect these aircraft to remain in service for at least another 20–30 years.

Patricia Marins

19,267 просмотров • 5 месяцев назад

This Chinese mathematician earned $10,000 a month inventing the hardest problems to train Neural Networks through Scale AI. Today his income dropped to zero. All the solutions are now generated by the model itself. He used to just hold the problem in his head and spell it out in plain text. His work is pure intellect. An expert in higher mathematics, he made his money hand-crafting the trickiest puzzles to test and train neural networks via RLHF. The bastion of "human" logic rested entirely on him, on people with PhDs who knew how to invent the problem. The collapse is simple. The shift to RLAIF and synthetic data. The model plays against itself, builds trees of logical inference, and solves deeper than a human can even invent the problem. No PhD data engineers, no hand-written prompt-completion examples, no manual grading. Just the model, search algorithms, and Chain of Thought. Ready-made "smart human-time" still sells on the market for many times more. His old rate was $50–100 per problem. The internal "mini-app" was written by the model too. Inside there's no pretty shell, just bare logic with exact steps: input: the problem statement inference tree: thousands of branches per second check: every step verifies itself output: a proof a human never had time to invent And here is what the whole setup looked like. He no longer needs to write an example by hand. He gave the model a direct instruction in human words, without a single formal term: "solve the problem yourself and grade yourself yourself" That's it. After that the algorithm found the solution, checked it, and trained on its own result, with no human. → the contractor got $50–100 per problem written → from 5,000 to 10,000 a month → now that income is annulled → a query to a math LLM costs 1–5 cents → a quant or an actuary runs 150,000–250,000 a year → the margin for whoever packages this into an agent is nearly 100% In the author's own words: "I'm no longer able to invent a problem the machine can't solve. The examiner became dumber than the one he's examining." But honestly, he admits the crude mistake himself, and it's not in the math, it's in the positioning. He tied his income to selling "smart human-time", to crafting formulas by hand. As long as he sells formulas, he's left behind. The machine computes faster than he can invent the problem. He names the right move himself: the role shifts from "intellectual craftsman" to "systems architect." Then he doesn't sell his time, he manages compute, packaging that same LLM into an autonomous agent that runs 24/7. Out of everything I've seen this year about the disappearance of intellectual professions, this is the most honest example: $50 per problem zeroed out to 1 cent per query, a doctor of science losing to a search algorithm, one problem stated in human words instead of a hand-written dataset, and right away an out-loud admission of the wrong business model. The barrier to entry in higher mathematics just dropped to the level of "describe the task in words." The only question is who'll be the first to stop selling their time and start managing the machine's compute.

This Chinese mathematician earned $10,000 a month inventing the hardest problems to train Neural Networks through Scale AI. Today his income dropped to zero. All the solutions are now generated by the model itself. He used to just hold the problem in his head and spell it out in plain text. His work is pure intellect. An expert in higher mathematics, he made his money hand-crafting the trickiest puzzles to test and train neural networks via RLHF. The bastion of "human" logic rested entirely on him, on people with PhDs who knew how to invent the problem. The collapse is simple. The shift to RLAIF and synthetic data. The model plays against itself, builds trees of logical inference, and solves deeper than a human can even invent the problem. No PhD data engineers, no hand-written prompt-completion examples, no manual grading. Just the model, search algorithms, and Chain of Thought. Ready-made "smart human-time" still sells on the market for many times more. His old rate was $50–100 per problem. The internal "mini-app" was written by the model too. Inside there's no pretty shell, just bare logic with exact steps: input: the problem statement inference tree: thousands of branches per second check: every step verifies itself output: a proof a human never had time to invent And here is what the whole setup looked like. He no longer needs to write an example by hand. He gave the model a direct instruction in human words, without a single formal term: "solve the problem yourself and grade yourself yourself" That's it. After that the algorithm found the solution, checked it, and trained on its own result, with no human. → the contractor got $50–100 per problem written → from 5,000 to 10,000 a month → now that income is annulled → a query to a math LLM costs 1–5 cents → a quant or an actuary runs 150,000–250,000 a year → the margin for whoever packages this into an agent is nearly 100% In the author's own words: "I'm no longer able to invent a problem the machine can't solve. The examiner became dumber than the one he's examining." But honestly, he admits the crude mistake himself, and it's not in the math, it's in the positioning. He tied his income to selling "smart human-time", to crafting formulas by hand. As long as he sells formulas, he's left behind. The machine computes faster than he can invent the problem. He names the right move himself: the role shifts from "intellectual craftsman" to "systems architect." Then he doesn't sell his time, he manages compute, packaging that same LLM into an autonomous agent that runs 24/7. Out of everything I've seen this year about the disappearance of intellectual professions, this is the most honest example: $50 per problem zeroed out to 1 cent per query, a doctor of science losing to a search algorithm, one problem stated in human words instead of a hand-written dataset, and right away an out-loud admission of the wrong business model. The barrier to entry in higher mathematics just dropped to the level of "describe the task in words." The only question is who'll be the first to stop selling their time and start managing the machine's compute.

Blaze

49,109 просмотров • 1 месяц назад

Most AI assistants can now generate impressive responses. Very few can actually move work forward. At Glean, we’ve been focused on a different question: what would it take for every employee to have an expert, agentic coworker that understands their business, their role, and their day—and can safely take action across it? Today we’re introducing the latest generation of Glean Assistant: a Work AI partner grounded in your enterprise context with new capabilities like: • Real-time voice, so people can talk to Assistant like a colleague. Get briefed on an account on the way to a meeting, ask follow-up questions, and line up next steps without touching a keyboard. • On-brand slide generation and in canvas, so teams can go from an idea and a few source docs to a polished, brand-aligned deck or message in minutes. • Agent sandboxes and 100+ enterprise actions, so Assistant can analyze large datasets and coordinate work across tools like Salesforce, Jira, Confluence, GitHub, Google Calendar, Asana, and Canva—helping teams move directly from insight to execution, with guardrails in place. • Visible, editable personal graph and proactive agents, so every employee can see and control the context Assistant uses to plan their day, surface action items, and summarize what they accomplished each week. Customers are already seeing nearly two hours saved per employee per day because their teams can quickly surface the right information, generate what they need, and move work forward without breaking focus. We’ll be sharing what this looks like in practice at Glean:LIVE – Context in Action today at 10 AM PT. I’d love for you to join us. Watch here:

Arvind Jain

15,524 просмотров • 5 месяцев назад

I'm running Llama 4 Maverick at 620 t/s! I'm living in the future! Honestly, a large language model running this fast is something straight out of a sci-fi movie. Speeds like this will enable a whole new world of applications that aren't possible today. For reference, GPT-4o, which is probably the most popular OpenAI model, runs between 60 and 110 t/s. The secret here: I'm not running AI at Meta's Llama 4 Maverick on a GPU. I'm using the SambaNova Cloud (my sponsor) and their custom SN40L chips. They are optimized from the ground up for running AI workflows. Right now, SambaNova Cloud runs DeepSeek, Qwen, Whisper, and the entire family of Llama models on these chips. You can check the speed of each of these models using SambaNova Cloud's Playground (see the attached video). It's completely free, and that's how I'm measuring their speeds. For example, I also tried DeepSeek R1 (the latest version from May) and, oh boy! DeepSeek R1 is a huge 671B parameter model. It's probably the best open reasoning model in the world, and it runs at 140 tokens per second! !!! Inference time on an SN40L is night and day from what you'll get from a GPU. Here is why this is big: If you are running an agentic workflow that uses multiple models simultaneously on a GPU, it will need to swap models in and out of memory (because not every model fits). A single SNL40 chip can simultaneously hold over 100 models (trillions of parameters) in memory. If you are using open models, try the SambaCloud API to see what lightning speed looks like. Here is how: 1. Create a free account at: 2. Check the QuickStart guide: If you try the playground, check the speed you're getting with Llama 4 and DeepSeek, and post the results below. I've seen much higher numbers than I posted here, so I'm curious to see whether geography affects the speed.

I'm running Llama 4 Maverick at 620 t/s! I'm living in the future! Honestly, a large language model running this fast is something straight out of a sci-fi movie. Speeds like this will enable a whole new world of applications that aren't possible today. For reference, GPT-4o, which is probably the most popular OpenAI model, runs between 60 and 110 t/s. The secret here: I'm not running AI at Meta's Llama 4 Maverick on a GPU. I'm using the SambaNova Cloud (my sponsor) and their custom SN40L chips. They are optimized from the ground up for running AI workflows. Right now, SambaNova Cloud runs DeepSeek, Qwen, Whisper, and the entire family of Llama models on these chips. You can check the speed of each of these models using SambaNova Cloud's Playground (see the attached video). It's completely free, and that's how I'm measuring their speeds. For example, I also tried DeepSeek R1 (the latest version from May) and, oh boy! DeepSeek R1 is a huge 671B parameter model. It's probably the best open reasoning model in the world, and it runs at 140 tokens per second! !!! Inference time on an SN40L is night and day from what you'll get from a GPU. Here is why this is big: If you are running an agentic workflow that uses multiple models simultaneously on a GPU, it will need to swap models in and out of memory (because not every model fits). A single SNL40 chip can simultaneously hold over 100 models (trillions of parameters) in memory. If you are using open models, try the SambaCloud API to see what lightning speed looks like. Here is how: 1. Create a free account at: 2. Check the QuickStart guide: If you try the playground, check the speed you're getting with Llama 4 and DeepSeek, and post the results below. I've seen much higher numbers than I posted here, so I'm curious to see whether geography affects the speed.

Santiago

34,148 просмотров • 1 год назад