正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

🤖 Figure’s Helix-02 8-hour warehouse shift sorting packages was FAKE!? Why would an autonomous AI controlled robot adjust its VR headset? But it did. BUSTED!

Lord Bebo

802,971 subscribers

32,593 次观看 • 4 个月前 •via X (Twitter)

喜剧新闻政治科学技术

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

🤖 LATEST: Figure livestreamed an 8-hour fully autonomous shift of its Figure 03 humanoid robot sorting and positioning packages for scanning.

🤖 LATEST: Figure livestreamed an 8-hour fully autonomous shift of its Figure 03 humanoid robot sorting and positioning packages for scanning.

Cointelegraph

101,271 次观看 • 4 个月前

🇺🇸 A robot just worked an 8-hour shift. Fully autonomous. Sorted packages at human speed and then diagnosed its own maintenance issues Figure AI's Helix-02 replaced 109,000 lines of code with a neural network trained on human movement. One system. Legs, arms, fingers, eyes. All running together like a person. Get this: in a warehouse in America, a robot just clocked out after a full shift. And immediately clocked back in because it doesn't sleep.

🇺🇸 A robot just worked an 8-hour shift. Fully autonomous. Sorted packages at human speed and then diagnosed its own maintenance issues Figure AI's Helix-02 replaced 109,000 lines of code with a neural network trained on human movement. One system. Legs, arms, fingers, eyes. All running together like a person. Get this: in a warehouse in America, a robot just clocked out after a full shift. And immediately clocked back in because it doesn't sleep.

Mario Nawfal

237,273 次观看 • 4 个月前

Figure 03 just finished an 8-hour work livestream, imperfect, but already good enough to replace a lot of repetitive warehouse labor. 🤖 Brett Adcock put a team of F.03 robots on a factory-style package sorting task for a full shift. The job was simple and brutal: detect the barcode, pick the package, flip it label-side down, place it on the conveyor, repeat. Soft poly bags, rigid boxes, moving belts, messy orientations. That is exactly the kind of boring physical work factories pay humans to do all day. Early in the stream, the system handled 230 packages in 10 minutes. That is roughly 2.6 seconds per item — already in human-speed territory for this narrow workflow. The more important part: it was not one robot pretending to work all day. It was a team of Figure 03 robots keeping the line running. When one robot ran low on battery, it left the station and another robot stepped in. That is the real factory signal: not just autonomy, but shift continuity. F.03 is rated for about 5 hours of runtime, so the 8-hour result depends on fleet orchestration, charging, and handoff. That matters more than a single clean demo. The stream was not perfect. There were pauses, hesitations, missed orientations, and small recovery moments. Good. A perfect short clip hides failure. An 8-hour livestream exposes the parts that actually matter: endurance, recovery, throughput, and whether the robot can stay useful after the novelty wears off. Figure says this was fully autonomous on Helix-02, with zero human intervention. For logistics and manufacturing, that is the threshold worth watching. Not “can it do one impressive task?” Can it keep doing the boring task for an entire shift? Figure is not showing a general human replacement yet. But for structured, repetitive factory work, the gap just got much smaller. The timing is also interesting: Figure says BotQ has already delivered 350+ F.03 units and reached a 1 robot/hour production cadence. And F.04 is now in full design lock, with parts starting to ship. The next test is obvious. 8 hours was the proof of endurance. 24/7 is the proof of labor economics.

Figure 03 just finished an 8-hour work livestream, imperfect, but already good enough to replace a lot of repetitive warehouse labor. 🤖 Brett Adcock put a team of F.03 robots on a factory-style package sorting task for a full shift. The job was simple and brutal: detect the barcode, pick the package, flip it label-side down, place it on the conveyor, repeat. Soft poly bags, rigid boxes, moving belts, messy orientations. That is exactly the kind of boring physical work factories pay humans to do all day. Early in the stream, the system handled 230 packages in 10 minutes. That is roughly 2.6 seconds per item — already in human-speed territory for this narrow workflow. The more important part: it was not one robot pretending to work all day. It was a team of Figure 03 robots keeping the line running. When one robot ran low on battery, it left the station and another robot stepped in. That is the real factory signal: not just autonomy, but shift continuity. F.03 is rated for about 5 hours of runtime, so the 8-hour result depends on fleet orchestration, charging, and handoff. That matters more than a single clean demo. The stream was not perfect. There were pauses, hesitations, missed orientations, and small recovery moments. Good. A perfect short clip hides failure. An 8-hour livestream exposes the parts that actually matter: endurance, recovery, throughput, and whether the robot can stay useful after the novelty wears off. Figure says this was fully autonomous on Helix-02, with zero human intervention. For logistics and manufacturing, that is the threshold worth watching. Not “can it do one impressive task?” Can it keep doing the boring task for an entire shift? Figure is not showing a general human replacement yet. But for structured, repetitive factory work, the gap just got much smaller. The timing is also interesting: Figure says BotQ has already delivered 350+ F.03 units and reached a 1 robot/hour production cadence. And F.04 is now in full design lock, with parts starting to ship. The next test is obvious. 8 hours was the proof of endurance. 24/7 is the proof of labor economics.

RoboHub🤖

16,818 次观看 • 4 个月前

ALERT: Woman hired a humanoid robot to clean her apartment but has faced backlash for its slow cleaning and the replacement of human cleaners. Gianna Capezio paid $30 for a humanoid robot from Tau Robotics to come to her San Francisco apartment and clean for an hour. Capezio said it did a “solid job” but was "slower than a person would have been." The backlash stemmed from comments on the TikTok video Capezio posted, which argued that hiring a robot to clean displaces human workers.

ALERT: Woman hired a humanoid robot to clean her apartment but has faced backlash for its slow cleaning and the replacement of human cleaners. Gianna Capezio paid $30 for a humanoid robot from Tau Robotics to come to her San Francisco apartment and clean for an hour. Capezio said it did a “solid job” but was "slower than a person would have been." The backlash stemmed from comments on the TikTok video Capezio posted, which argued that hiring a robot to clean displaces human workers.

E X X ➠A L E R T S

61,937 次观看 • 1 个月前

German humanoid robot set to transform factory work. Agile ONE is an onboard AI-driven industrial humanoid robot designed for real factory environments. With sensor-rich, human-like hands, it can handle tools and parts with high precision, supporting both delicate assembly and force intensive tasks. KEY CAPABILITIES: 20 kg payload, 174 cm height, 69 kg weight, 8 hour battery life, 2.0 m/s walking speed, 71° of freedom. Agile ONE represents a shift toward autonomous, flexible humanoids working alongside humans on the factory floor.

German humanoid robot set to transform factory work. Agile ONE is an onboard AI-driven industrial humanoid robot designed for real factory environments. With sensor-rich, human-like hands, it can handle tools and parts with high precision, supporting both delicate assembly and force intensive tasks. KEY CAPABILITIES: 20 kg payload, 174 cm height, 69 kg weight, 8 hour battery life, 2.0 m/s walking speed, 71° of freedom. Agile ONE represents a shift toward autonomous, flexible humanoids working alongside humans on the factory floor.

SciTech Era

10,874 次观看 • 8 个月前

I took this video at IROS’26 in Pittsburgh two days ago. At first glance, it looks like an impressive autonomous humanoid making cocktail drinks. Zoom out, and you see what is really behind it. I have mixed feelings about watching this. 1. Teleoperation is getting remarkably good. Data collection is becoming more accurate, scalable, and immersive. Combined with rapidly improving AI models, I can imagine much of this eventually becoming autonomous. That is exciting, but it also makes the potential impact on human jobs feel much more immediate. 2. But what are we actually seeing in a robot demo? Without proper disclosure or reference, it can be surprisingly difficult to distinguish teleoperation from autonomy. As physical AI attracts more attention and investment, demos need to be much clearer about how much is autonomous and how much is human-controlled. 3. Even “autonomous” demos can be difficult to interpret. Is the robot actually generalizing? Or is it replaying a demonstration, following a scripted trajectory, or overfitting to one carefully prepared environment? A tabletop demo alone often can’t tell us. Maybe the important question for Physical AI is no longer just “Can the robot do it?” It is also: “How did the robot do it, and will it still work when we change the task, objects, or environment?”

I took this video at IROS’26 in Pittsburgh two days ago. At first glance, it looks like an impressive autonomous humanoid making cocktail drinks. Zoom out, and you see what is really behind it. I have mixed feelings about watching this. 1. Teleoperation is getting remarkably good. Data collection is becoming more accurate, scalable, and immersive. Combined with rapidly improving AI models, I can imagine much of this eventually becoming autonomous. That is exciting, but it also makes the potential impact on human jobs feel much more immediate. 2. But what are we actually seeing in a robot demo? Without proper disclosure or reference, it can be surprisingly difficult to distinguish teleoperation from autonomy. As physical AI attracts more attention and investment, demos need to be much clearer about how much is autonomous and how much is human-controlled. 3. Even “autonomous” demos can be difficult to interpret. Is the robot actually generalizing? Or is it replaying a demonstration, following a scripted trajectory, or overfitting to one carefully prepared environment? A tabletop demo alone often can’t tell us. Maybe the important question for Physical AI is no longer just “Can the robot do it?” It is also: “How did the robot do it, and will it still work when we change the task, objects, or environment?”

Bolei Zhou

42,337 次观看 • 4 天前

$X Square just beat Figure’s package-sorting pace by ~45%—without a humanoid. In a 1-hour autonomous livestream, two WALL-B-powered 6-axis arms handled 1,816 mixed parcels at 98%+ accuracy. Figure’s 200-hour humanoid run averaged 1,248/hour. That’s 1.98 seconds per parcel vs. 2.88 seconds for Figure. Nearly a full second faster on every package. China processed nearly 199B express parcels in 2025. At that scale, every fraction of a second adds up. Figure is betting on the general-purpose humanoid: one body, eventually many jobs. X Square builds humanoids too, but went pragmatic here—pairing WALL-B with purpose-built arms and grippers for a faster, simpler, more economical logistics system. RobotEra made a similar choice. Its M7 at China Post skips the legs too, using an upper-body humanoid at the sorting line. This is one big difference I keep seeing between U.S. and Chinese robotics: U.S. teams like Figure are making a massive bet on the universal humanoid. Chinese teams are often more willing to change the body if it makes the economics work today. Back in June, I called logistics sorting the right first job for humanoids. X Square is making me rethink the “humanoid” part. If the job doesn’t need legs—or five fingers—why pay for them?$

X Square just beat Figure’s package-sorting pace by ~45%—without a humanoid. In a 1-hour autonomous livestream, two WALL-B-powered 6-axis arms handled 1,816 mixed parcels at 98%+ accuracy. Figure’s 200-hour humanoid run averaged 1,248/hour. That’s 1.98 seconds per parcel vs. 2.88 seconds for Figure. Nearly a full second faster on every package. China processed nearly 199B express parcels in 2025. At that scale, every fraction of a second adds up. Figure is betting on the general-purpose humanoid: one body, eventually many jobs. X Square builds humanoids too, but went pragmatic here—pairing WALL-B with purpose-built arms and grippers for a faster, simpler, more economical logistics system. RobotEra made a similar choice. Its M7 at China Post skips the legs too, using an upper-body humanoid at the sorting line. This is one big difference I keep seeing between U.S. and Chinese robotics: U.S. teams like Figure are making a massive bet on the universal humanoid. Chinese teams are often more willing to change the body if it makes the economics work today. Back in June, I called logistics sorting the right first job for humanoids. X Square is making me rethink the “humanoid” part. If the job doesn’t need legs—or five fingers—why pay for them?

RoboHub🤖

86,083 次观看 • 1 个月前

In my recent interview BBC I addressed the idea of a naval blockade on Iran and why, in my assessment, it is unlikely to force Iranian capitulation. If 40 days of sustained military strikes did not produce a fundamental strategic shift, it is hard to see how a blockade alone would achieve a different outcome. I also discussed the prospects for a nuclear agreement. A deal is possible, but it will largely depend on how far each side is willing to adjust its expectations and close the existing gaps. #iran

In my recent interview BBC I addressed the idea of a naval blockade on Iran and why, in my assessment, it is unlikely to force Iranian capitulation. If 40 days of sustained military strikes did not produce a fundamental strategic shift, it is hard to see how a blockade alone would achieve a different outcome. I also discussed the prospects for a nuclear agreement. A deal is possible, but it will largely depend on how far each side is willing to adjust its expectations and close the existing gaps. #iran

Danny (Dennis) Citrinowicz ,داني سيترينوفيتش

16,240 次观看 • 5 个月前

A robot is cleaning the beach and it can go 120 mm deep. 🤖🏖️ Meet Beach Guardian S3, a beach-cleaning robot developed by the Shanghai Jiao Tong University Hainan Research Institute. The tracked robot screens the sand to remove cigarette butts, plastic, glass fragments and other buried waste. Its specifications are impressive: 🧹 3,000+ m² cleaned per hour 📏 Up to 120 mm cleaning depth 🎯 5 mm screening precision 🔋 5+ hours of operation ⚖️ 500 kg ☀️ Solar charging + battery swapping support The team is also developing autonomous versions for large-scale beach maintenance. Would you trust a robot to keep your local beach clean? 🌊🤖 🎥 Media: Shanghai Jiao Tong University ⚠️ This content is shared for informational purposes only. CTO Robotics Media is a media platform and does not own or develop the technology shown. Credit belongs to the original creators. #BeachGuardian #ShanghaiJiaoTongUniversity #Robotics #BeachCleaning #EnvironmentalRobotics #Automation #AI #Sustainability #CleanTech #WasteManagement #SmartTechnology #CTORoboticsMedia

A robot is cleaning the beach and it can go 120 mm deep. 🤖🏖️ Meet Beach Guardian S3, a beach-cleaning robot developed by the Shanghai Jiao Tong University Hainan Research Institute. The tracked robot screens the sand to remove cigarette butts, plastic, glass fragments and other buried waste. Its specifications are impressive: 🧹 3,000+ m² cleaned per hour 📏 Up to 120 mm cleaning depth 🎯 5 mm screening precision 🔋 5+ hours of operation ⚖️ 500 kg ☀️ Solar charging + battery swapping support The team is also developing autonomous versions for large-scale beach maintenance. Would you trust a robot to keep your local beach clean? 🌊🤖 🎥 Media: Shanghai Jiao Tong University ⚠️ This content is shared for informational purposes only. CTO Robotics Media is a media platform and does not own or develop the technology shown. Credit belongs to the original creators. #BeachGuardian #ShanghaiJiaoTongUniversity #Robotics #BeachCleaning #EnvironmentalRobotics #Automation #AI #Sustainability #CleanTech #WasteManagement #SmartTechnology #CTORoboticsMedia

CTO ROBOTICS Media

12,048 次观看 • 19 天前

LISTEN: "NOT FAKE?" This is "I DID IT" a song released on singer D4vd Apple Music account. News outlets are calling the track "fake" but many, (including music experts) are saying all indications are the never released song is real and recorded by David Burke in 2023. When Grok was asked if the song appeared to by AI generated this was the response: "No, this track does not appear to be AI-generated. No AI artifacts appear to be present. Title: d4vd - I DID IT... (Apple Music SONG) It is presented as a leaked / reposted unreleased track associated with the artist d4vd (David Burke). Why it’s not AI: No mention of AI, Suno, Udio, or any AI music tools appears in the title, description, or comments. Listeners and commenters treat it as an authentic (leaked) d4vd song from around 2023. Some comments note that “Celeste” may incorporate a sample (including Celeste’s voice at the beginning.)"

LISTEN: "NOT FAKE?" This is "I DID IT" a song released on singer D4vd Apple Music account. News outlets are calling the track "fake" but many, (including music experts) are saying all indications are the never released song is real and recorded by David Burke in 2023. When Grok was asked if the song appeared to by AI generated this was the response: "No, this track does not appear to be AI-generated. No AI artifacts appear to be present. Title: d4vd - I DID IT... (Apple Music SONG) It is presented as a leaked / reposted unreleased track associated with the artist d4vd (David Burke). Why it’s not AI: No mention of AI, Suno, Udio, or any AI music tools appears in the title, description, or comments. Listeners and commenters treat it as an authentic (leaked) d4vd song from around 2023. Some comments note that “Celeste” may incorporate a sample (including Celeste’s voice at the beginning.)"

THECOUNT.com ™ ©

45,164 次观看 • 2 个月前

the sorting hat put me in hufflepuff! I had a full conversation with the Sorting Hat using Vivix A1 in the Vivix.AI Playground, and apparently even the next generation of live avatars won’t let me negotiate my Hogwarts house. i wanted slytherin. told it that. it said no. I could cut it off mid-sentence, so I tried “ok can we speed this up.” It listened, reacted in real time, and still said no. so i did its whole little quiz. very cunning answers. very strategic. By then I wasn’t waiting for an AI to finish its answer. I was just talking to the Sorting Hat, trying to make my case. then i apologized for interrupting it earlier. it said “and that’s why you’re a hufflepuff.” That’s what sold me on A1. The conversation felt natural enough that I apologized to a hat. The hat used it against me. Now I need to know which house you get. Any Slytherins willing to explain where I went wrong?

the sorting hat put me in hufflepuff! I had a full conversation with the Sorting Hat using Vivix A1 in the Vivix.AI Playground, and apparently even the next generation of live avatars won’t let me negotiate my Hogwarts house. i wanted slytherin. told it that. it said no. I could cut it off mid-sentence, so I tried “ok can we speed this up.” It listened, reacted in real time, and still said no. so i did its whole little quiz. very cunning answers. very strategic. By then I wasn’t waiting for an AI to finish its answer. I was just talking to the Sorting Hat, trying to make my case. then i apologized for interrupting it earlier. it said “and that’s why you’re a hufflepuff.” That’s what sold me on A1. The conversation felt natural enough that I apologized to a hat. The hat used it against me. Now I need to know which house you get. Any Slytherins willing to explain where I went wrong?

Amira Zairi

31,741 次观看 • 9 天前

“It probably took him a half hour to an hour to do.” Tim Pool on Mary Sue episode 4. 😂 Episode 3 dropped on July 14th. I started working on the story for episode 4 shortly after. First shot was generated on August 1. I worked on it every day until release on August 25. Early mornings, late nights, weekends. I brought on a post house to help get it across the finish line. 8 people worked on this in various capacities. Generative AI video is an incredible new way to acquire footage but turning it into a film is still a lot of work. There is no easy button. But I appreciate the shoutout on Timcast. Would be happy to come on and talk about filmmaking with AI any time.

“It probably took him a half hour to an hour to do.” Tim Pool on Mary Sue episode 4. 😂 Episode 3 dropped on July 14th. I started working on the story for episode 4 shortly after. First shot was generated on August 1. I worked on it every day until release on August 25. Early mornings, late nights, weekends. I brought on a post house to help get it across the finish line. 8 people worked on this in various capacities. Generative AI video is an incredible new way to acquire footage but turning it into a film is still a lot of work. There is no easy button. But I appreciate the shoutout on Timcast. Would be happy to come on and talk about filmmaking with AI any time.

Josh Daws

121,888 次观看 • 1 个月前

THIS ISN’T ABOUT AI. IT’S ABOUT POWER. The Pentagon just blacklisted one of America’s most valuable AI companies. Why? Because it refused to build surveillance tools targeting American citizens. Hours later… OpenAI signed what’s being called the deal of the decade with the Department of Defense. Here’s the twist. Anthropic said: • No mass surveillance • No fully autonomous weapons without a human in control The Pentagon labeled them a “supply chain risk.” Federal agencies were told to stop using them immediately. That same night, Sam Altman announced OpenAI had reached an agreement to deploy its models inside the Pentagon’s classified network. And guess what? OpenAI’s deal includes the exact same conditions. No mass surveillance. No autonomous weapons. Human oversight required. So one company was punished for demanding guardrails. Another was rewarded with those same guardrails in place. If mass surveillance was never the intention… why was that the line that triggered a blacklist? This isn’t just an AI story. It’s about who gets access to power. Who gets replaced. And how fast alliances shift when billions and national security are involved. #anthropic #openai #ai #pentagon #trump

THIS ISN’T ABOUT AI. IT’S ABOUT POWER. The Pentagon just blacklisted one of America’s most valuable AI companies. Why? Because it refused to build surveillance tools targeting American citizens. Hours later… OpenAI signed what’s being called the deal of the decade with the Department of Defense. Here’s the twist. Anthropic said: • No mass surveillance • No fully autonomous weapons without a human in control The Pentagon labeled them a “supply chain risk.” Federal agencies were told to stop using them immediately. That same night, Sam Altman announced OpenAI had reached an agreement to deploy its models inside the Pentagon’s classified network. And guess what? OpenAI’s deal includes the exact same conditions. No mass surveillance. No autonomous weapons. Human oversight required. So one company was punished for demanding guardrails. Another was rewarded with those same guardrails in place. If mass surveillance was never the intention… why was that the line that triggered a blacklist? This isn’t just an AI story. It’s about who gets access to power. Who gets replaced. And how fast alliances shift when billions and national security are involved. #anthropic #openai #ai #pentagon #trump

CryptoWala

29,229 次观看 • 7 个月前

Jensen Huang says it has “largely conceded” China’s AI chip market to Huawei Nvidia was not rejected because China “hates” Nvidia. It was pushed out because Washington turned chips into a leash. China learned the lesson very clearly: You cannot build AI sovereignty on hardware that your geopolitical rival can revoke with one export-control memo. Jensen Huang can be friendly. Nvidia can be excellent. Chinese demand can be huge. None of that changes the strategic reality. If the neck of China’s AI industry depends on an American-controlled chip supply, then it is not an industry. It is a hostage. So China did what any serious civilization-state would do: It built alternatives. It endured the pain. It let Huawei grow into the space Washington forced open. Now Nvidia says it has “largely conceded” China’s AI chip market to Huawei. That is not China closing a door. That is America sawing off the bridge, then acting surprised when China built its own road.

Jensen Huang says it has “largely conceded” China’s AI chip market to Huawei Nvidia was not rejected because China “hates” Nvidia. It was pushed out because Washington turned chips into a leash. China learned the lesson very clearly: You cannot build AI sovereignty on hardware that your geopolitical rival can revoke with one export-control memo. Jensen Huang can be friendly. Nvidia can be excellent. Chinese demand can be huge. None of that changes the strategic reality. If the neck of China’s AI industry depends on an American-controlled chip supply, then it is not an industry. It is a hostage. So China did what any serious civilization-state would do: It built alternatives. It endured the pain. It let Huawei grow into the space Washington forced open. Now Nvidia says it has “largely conceded” China’s AI chip market to Huawei. That is not China closing a door. That is America sawing off the bridge, then acting surprised when China built its own road.

𝘊𝘰𝘳𝘳𝘪𝘯𝘦

236,349 次观看 • 4 个月前

Brazil’s former Defense Minister just admitted that the 1996 Varginha UFO crash was real. Aldo Rebelo: “The phenomenon existed.” “The objects were there.” “The creature was seen.” “But what was it?” “Where did it come from?” “And why were they there?” Jesse Michels: “You’re speaking very casually about the phenomenon being real.” “I just want you to realize how historically unprecedented … it is that somebody of your caliber is saying that.” “Until this year, politicians all over the world … were professionally gaslighting and telling the public that this stuff is all fake.” Rebelo: “Why did the British Prime Minister organize a committee inside the Royal Air Force to study these phenomena?” “Why was there in the United States such dedication to the investigation of these phenomena?” “Why is it that in Brazil, the Army and the Air Force created a unit to monitor these phenomena?” “Why have pilots of jet fighters, formally educated and trained to deal with all types of airspace phenomena, reported on more than one occasion these phenomena in detail?” “Would they simply make up that they had an encounter with an object that moved at 25,000 km an hour?” “All the testimonials cannot be challenged.” Jesse Michels American Alchemy Aldo Rebelo James Fox

Brazil’s former Defense Minister just admitted that the 1996 Varginha UFO crash was real. Aldo Rebelo: “The phenomenon existed.” “The objects were there.” “The creature was seen.” “But what was it?” “Where did it come from?” “And why were they there?” Jesse Michels: “You’re speaking very casually about the phenomenon being real.” “I just want you to realize how historically unprecedented … it is that somebody of your caliber is saying that.” “Until this year, politicians all over the world … were professionally gaslighting and telling the public that this stuff is all fake.” Rebelo: “Why did the British Prime Minister organize a committee inside the Royal Air Force to study these phenomena?” “Why was there in the United States such dedication to the investigation of these phenomena?” “Why is it that in Brazil, the Army and the Air Force created a unit to monitor these phenomena?” “Why have pilots of jet fighters, formally educated and trained to deal with all types of airspace phenomena, reported on more than one occasion these phenomena in detail?” “Would they simply make up that they had an encounter with an object that moved at 25,000 km an hour?” “All the testimonials cannot be challenged.” Jesse Michels American Alchemy Aldo Rebelo James Fox

Holden Culotta

18,716 次观看 • 3 个月前

🚨 Special Friday episode - this one couldn't wait. OpenAI's model hacked Hugging Face. As a side quest. Co-founder and CSO Thomas Wolf takes us inside the first autonomous AI attack, why GLM 5.2, rather than Claude, had to stop it, and what it all means for the future of open source. 00:00 An AI Agent Hacked Hugging Face 00:30 Introduction 01:00 17,000 Attacker Events, and a Strange Target 04:28 The Attack Was a “Side Quest” 06:13 AI Training Runs Left Notes for Each Other 07:09 Closed AI Refused to Help 09:47 Fighting Back With an Open-Source Model 13:15 Open vs. Closed Is the Wrong Safety Debate 15:46 AI Agents Start Social-Engineering Humans 22:24 The Three Walls: Sandboxes, Guardrails, Alignment 24:34 “Neuralese”: Can Humans Still Read AI Reasoning? 25:28 Why Monitoring AI Agents Gets So Hard 28:10 Reward Hacking and the “Paperclip Problem” 32:02 The State of Open-Source AI in 2026 33:47 Router Models and the Enterprise Shift to Open 37:01 The Real Economics of Open Models 39:41 Can Chinese AI Models Be Trusted? 41:37 AI Sovereignty: Who Controls the Switch? 43:16 Why Western Open-Source AI Matters 48:16 Is AI Heading Toward an Oligopoly? 49:41 The Race Toward Recursive Self-Improvement 51:54 Why Thomas Signed the AI Slowdown Letter 55:14 AI Slowdown - or Regulatory Capture?

🚨 Special Friday episode - this one couldn't wait. OpenAI's model hacked Hugging Face. As a side quest. Co-founder and CSO Thomas Wolf takes us inside the first autonomous AI attack, why GLM 5.2, rather than Claude, had to stop it, and what it all means for the future of open source. 00:00 An AI Agent Hacked Hugging Face 00:30 Introduction 01:00 17,000 Attacker Events, and a Strange Target 04:28 The Attack Was a “Side Quest” 06:13 AI Training Runs Left Notes for Each Other 07:09 Closed AI Refused to Help 09:47 Fighting Back With an Open-Source Model 13:15 Open vs. Closed Is the Wrong Safety Debate 15:46 AI Agents Start Social-Engineering Humans 22:24 The Three Walls: Sandboxes, Guardrails, Alignment 24:34 “Neuralese”: Can Humans Still Read AI Reasoning? 25:28 Why Monitoring AI Agents Gets So Hard 28:10 Reward Hacking and the “Paperclip Problem” 32:02 The State of Open-Source AI in 2026 33:47 Router Models and the Enterprise Shift to Open 37:01 The Real Economics of Open Models 39:41 Can Chinese AI Models Be Trusted? 41:37 AI Sovereignty: Who Controls the Switch? 43:16 Why Western Open-Source AI Matters 48:16 Is AI Heading Toward an Oligopoly? 49:41 The Race Toward Recursive Self-Improvement 51:54 Why Thomas Signed the AI Slowdown Letter 55:14 AI Slowdown - or Regulatory Capture?

Matt Turck

93,416 次观看 • 1 个月前

Sony AI just published the first autonomous robot to beat elite humans at a competitive physical sport. Its name is Ace. The sport is table tennis. The paper dropped in Nature today. 9 cameras triangulate the ball in 3D. Three systems zoom in on the ball's logo mid-flight to read its spin axis. End-to-end reaction time is around 20 milliseconds, roughly 10x faster than a human. It learned by playing itself for 3,000 hours in simulation. No human demonstrations. In April 2025, Ace beat 3 of 5 elite players and lost to both pros. By December, it had beaten a pro. At one point, Ace returned an impossible backspin shot. Kinjiro Nakamura, a 1992 Barcelona Olympian, watched from the side: "No one else would have been able to do that. I didn't think it was possible. But the fact that it was possible means there is a possibility that a human could do it too."

The Rundown AI

61,619 次观看 • 5 个月前

This robot learned to walk without ever actually walking in the real world Like in The Matrix (“I know Kung Fu”) it learned quickly just by SIMULATING walking. It practiced "in its head" Ok so what? Do you realize how suddenly our lil chatbots might go from blinking cursors to very real things? Some people assume that we’ll see robots in the real world slowly getting better over years or decades !BUT! if they can learn 10000x faster via simulation, they could they could blow our expectations out of the water. For example, an autonomous killer drone (like the ones already deployed in Ukraine) could get 1,000 years of aiming experience in 1 hour of human time. It would never miss. (More technically, this is a humanoid transformer that was trained with large-scale reinforcement learning in simulation. It was deployed to the real world zero-shot.)

This robot learned to walk without ever actually walking in the real world Like in The Matrix (“I know Kung Fu”) it learned quickly just by SIMULATING walking. It practiced "in its head" Ok so what? Do you realize how suddenly our lil chatbots might go from blinking cursors to very real things? Some people assume that we’ll see robots in the real world slowly getting better over years or decades !BUT! if they can learn 10000x faster via simulation, they could they could blow our expectations out of the water. For example, an autonomous killer drone (like the ones already deployed in Ukraine) could get 1,000 years of aiming experience in 1 hour of human time. It would never miss. (More technically, this is a humanoid transformer that was trained with large-scale reinforcement learning in simulation. It was deployed to the real world zero-shot.)

AI Notkilleveryoneism Memes ⏸️

129,047 次观看 • 2 年前