Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

What we’re seeing: frontier robotics teams are rebuilding data pipelines from scratch because generic licensed datasets aren’t moving their models forward. The next wave won’t be about more data. It’ll be about precise, domain-specific data that actually improves performance. The bottleneck isn’t compute. It’s the right data. 🎥 Our... show more

Perle Labs

253,831 subscribers

30,624 Aufrufe • vor 7 Monaten •via X (Twitter)

Wissenschaft & Technologie

Anya Rossi• Live Now

Private livecam show

22 Kommentare

Profilbild von XR🦇

XR🦇vor 7 Monaten

No lies here Reason Perle Labs plan on making humans the center of AI!

Profilbild von CryptoManiaUA

CryptoManiaUAvor 7 Monaten

Totally agree. The bottleneck isn’t more scale — it’s higher signal. Domain-specific data is the real competitive edge.

Profilbild von Gabrielight

Gabrielightvor 7 Monaten

Gperle

Profilbild von NEERAJ KUMAR

NEERAJ KUMARvor 7 Monaten

gperle 🚀

Profilbild von Abhishek Singh

Abhishek Singhvor 7 Monaten

"This is such an important insight. 🚀 The future of robotics & AI isn’t about bigger datasets—it’s about better, domain-specific data that actually drives performance. Precision > volume. Huge takeaway from Multichain Day at EthereumDenver 👏"

Profilbild von Konigthegreit

Konigthegreitvor 7 Monaten

AI+robotics =🔥

Profilbild von monothiez 🐳

monothiez 🐳vor 7 Monaten

🙌🙌

Profilbild von Solix.signNFT

Solix.signNFTvor 7 Monaten

Yoh! I found this helpful

Profilbild von Quivex

Quivexvor 7 Monaten

data's overrated man, just can’t keep up with it

Profilbild von Zoroo | PoD 🦾

Zoroo | PoD 🦾vor 7 Monaten

Absolutely right — the data bottleneck in robotics isn't just volume; it's about domain-specific precision that actually moves benchmarks. Rebuilding pipelines from scratch makes sense when generic datasets fall short. Great insight from Arthur at Wrapped.

Profilbild von Muhammad Javed

Muhammad Javedvor 7 Monaten

When TGE

Profilbild von tanvir |M| iqbal

tanvir |M| iqbalvor 7 Monaten

frontier models don’t need more tokens, they need higher-signal, domain-specific data that actually shifts performance.

Profilbild von ge zu

ge zuvor 7 Monaten

Right, because this image can be found in a reverse image search, and the left and middle ones look unnatural to the eye, you can immediately tell they were AI-generated by the clouds and the colors of the hot air balloons

Profilbild von 𝓡•𝓞•𝓜•𝓘

𝓡•𝓞•𝓜•𝓘vor 7 Monaten

The future lies in verified and accurate data. Perle Labs is blazing this thorny path

Profilbild von Vk Choudhary

Vk Choudharyvor 7 Monaten

I've seen this play out, the right data always beats more data.

Profilbild von vorpal

vorpalvor 7 Monaten

Keep building 🫡

Profilbild von Rebecca Fox π²

Rebecca Fox π²vor 7 Monaten

Spot on. Compute is easy to rent, but high-fidelity, specialized data is the real moat. Quality over quantity is the only way to move the needle in robotics. 🤖🔥

Profilbild von Hakanoghlu.

Hakanoghlu.vor 7 Monaten

Check it out

Profilbild von nate.immutable

nate.immutablevor 7 Monaten

The bottleneck is quality data Gperle

Profilbild von Oluwafemi Daniel | 𝔽rAI

Oluwafemi Daniel | 𝔽rAIvor 7 Monaten

🔥

Profilbild von RAJツ

RAJツvor 7 Monaten

Data is a backbone for ai models and pErlebs AI now how to build a ai with help of correct data nice and best for upcoming AI for humanbeing..!

Profilbild von Najib Magaji

Najib Magajivor 7 Monaten

Love it.

Ähnliche Videos

How close is too close? For too many families across Virginia, massive data center campuses are right across the street from their homes. Their concerns about noise, health, and their communities deserve to be heard. That’s why we’re listening. Join us tomorrow in Emporia for our next Data Center Listening Tour stop.

How close is too close? For too many families across Virginia, massive data center campuses are right across the street from their homes. Their concerns about noise, health, and their communities deserve to be heard. That’s why we’re listening. Join us tomorrow in Emporia for our next Data Center Listening Tour stop.

L. Louise Lucas

11,597 Aufrufe • vor 11 Tagen

We might be solving the wrong problem in robotics. That’s what this makes clear. UMI → Universal Manipulation Interface A simple $400 gripper that lets you teach robots by demonstration. You hold it like a tool. Show the task. The robot learns. No teleoperation. No expensive hardware. No robot-specific data. Stanford open-sourced everything → hardware, code, datasets. What stands out to me is the bottleneck. Not algorithms. Data. Teleoperation → ~35 demos/hour UMI → ~111 demos/hour And the data transfers across robots → UR5, Franka, others. The design is surprisingly practical: → GoPro fisheye lens (155° FOV) + mirrors for depth → SLAM + IMU for precise 6DoF tracking → latency matching for dynamic tasks → diffusion policies for multimodal actions Then it scales. Cheng Chi takes this further with Sunday Robotics (with Tony Zhao). A $200 glove → deployed in 500+ homes → ~10 million real-world interactions. Not lab data. Real human behavior. Their robot learns dishes, laundry, espresso → with zero robot-specific data. This is where the shift becomes obvious. From training robots in controlled environments → to learning directly from humans at scale So here’s the real question: Will robotics be unlocked by better models… or by unlocking data? #ArtificialIntelligence #Robotics #AI #Innovation #FutureOfWork

We might be solving the wrong problem in robotics. That’s what this makes clear. UMI → Universal Manipulation Interface A simple $400 gripper that lets you teach robots by demonstration. You hold it like a tool. Show the task. The robot learns. No teleoperation. No expensive hardware. No robot-specific data. Stanford open-sourced everything → hardware, code, datasets. What stands out to me is the bottleneck. Not algorithms. Data. Teleoperation → ~35 demos/hour UMI → ~111 demos/hour And the data transfers across robots → UR5, Franka, others. The design is surprisingly practical: → GoPro fisheye lens (155° FOV) + mirrors for depth → SLAM + IMU for precise 6DoF tracking → latency matching for dynamic tasks → diffusion policies for multimodal actions Then it scales. Cheng Chi takes this further with Sunday Robotics (with Tony Zhao). A $200 glove → deployed in 500+ homes → ~10 million real-world interactions. Not lab data. Real human behavior. Their robot learns dishes, laundry, espresso → with zero robot-specific data. This is where the shift becomes obvious. From training robots in controlled environments → to learning directly from humans at scale So here’s the real question: Will robotics be unlocked by better models… or by unlocking data? #ArtificialIntelligence #Robotics #AI #Innovation #FutureOfWork

Pascal Bornet

186,783 Aufrufe • vor 6 Monaten

Jason Ma reveals the first true scaling law in robotics: robots trained on 1 million hours of first-person human video predictably get better without ever seeing robot data "Yesterday we announced our new flagship robot foundation model called Dyna-2. It's very significant for the entire field for many different reasons. First of all, it's the first robot foundation model trained on at least one million hours of data. That in itself is a very challenging infrastructure challenge." "What we demonstrated is that even if the one million hours of data is entirely first-person video of humans doing manipulation tasks, we actually see scaling law for transferring to robot embodiments that the model has never seen before." "By training on more human data, we actually see predictable performance improvement on robots. That's a big deal because the biggest challenge facing robot foundation models is that we don't have enough data." "Unlike language models, there are just not enough robotics data out there, and it's very difficult to collect these robot data. Only by passively observing humans doing things can we scale this kind of data, and this is the first time we showed there is a lot of improvement that can transfer." Jason Ma Dyna Robotics

Jason Ma reveals the first true scaling law in robotics: robots trained on 1 million hours of first-person human video predictably get better without ever seeing robot data "Yesterday we announced our new flagship robot foundation model called Dyna-2. It's very significant for the entire field for many different reasons. First of all, it's the first robot foundation model trained on at least one million hours of data. That in itself is a very challenging infrastructure challenge." "What we demonstrated is that even if the one million hours of data is entirely first-person video of humans doing manipulation tasks, we actually see scaling law for transferring to robot embodiments that the model has never seen before." "By training on more human data, we actually see predictable performance improvement on robots. That's a big deal because the biggest challenge facing robot foundation models is that we don't have enough data." "Unlike language models, there are just not enough robotics data out there, and it's very difficult to collect these robot data. Only by passively observing humans doing things can we scale this kind of data, and this is the first time we showed there is a lot of improvement that can transfer." Jason Ma Dyna Robotics

MTS

17,740 Aufrufe • vor 1 Monat

Everyone is talking about AI models, but almost no one is talking about data ownership. For years, centralized systems have dominated data pipelines, opaque processes, hidden biases, and zero contributor visibility. We call it innovation…, but in reality, it’s a black box. And that’s exactly how we ended up facing model degradation risks today. Because here’s the truth: AI is only as good as the data it learns from. That’s why Sovereign AI Data matters. To me, it means control, traceability, and accountability. Perle Labs is approaching this differently: ◈ On-chain provenance → Every data point is verifiable ◈ Expert-driven input → Quality over quantity ◈ Reputation systems → Trust becomes measurable This isn’t just a technical upgrade. It’s a paradigm shift. Instead of: Anonymous contributors ➠Unverified datasets ➠Centralized ownership ➠We move toward: Transparent data flows, Credible expertise and Distributed value creation If AI is the brain, then data is the memory. And right now most systems are learning from memory, which they can’t verify. Perle is solving that. While others focus on scaling models… This is about fixing the foundation they stand on. And foundations decide everything. #PerleAI #ToPerle participating in Perle Labs community campaign

Everyone is talking about AI models, but almost no one is talking about data ownership. For years, centralized systems have dominated data pipelines, opaque processes, hidden biases, and zero contributor visibility. We call it innovation…, but in reality, it’s a black box. And that’s exactly how we ended up facing model degradation risks today. Because here’s the truth: AI is only as good as the data it learns from. That’s why Sovereign AI Data matters. To me, it means control, traceability, and accountability. Perle Labs is approaching this differently: ◈ On-chain provenance → Every data point is verifiable ◈ Expert-driven input → Quality over quantity ◈ Reputation systems → Trust becomes measurable This isn’t just a technical upgrade. It’s a paradigm shift. Instead of: Anonymous contributors ➠Unverified datasets ➠Centralized ownership ➠We move toward: Transparent data flows, Credible expertise and Distributed value creation If AI is the brain, then data is the memory. And right now most systems are learning from memory, which they can’t verify. Perle is solving that. While others focus on scaling models… This is about fixing the foundation they stand on. And foundations decide everything. #PerleAI #ToPerle participating in Perle Labs community campaign

Big_Wealth❤️‍🔥

11,486 Aufrufe • vor 6 Monaten

A lot of robot learning pipelines treat embodiment mismatch as a software problem. Collect data with one device, retarget the motion to another robot, and try to deal with the differences later. TwinDEX flips that idea around. This system is from a Chinese robotics company called X Square. Instead of starting with a generic data-collection device, the team co-designs a wearable three-finger end-effector and a robotic hand as two sides of the same learning system. They align the things that actually matter during manipulation: joint structure, link proportions, contact geometry, surface properties, visual appearance, sensing, and timing. That matters because dexterous manipulation is much more than matching hand poses. A few millimeters of contact error or a different force relationship can completely change how a task is executed. In the collection evaluation, TwinDEX delivered up to 5.3 times the effective throughput of on-robot teleoperation. It’s a useful reminder that scaling robot learning may not only be about bigger models and more data. Sometimes the hardware collecting the data is part of the data problem itself.

A lot of robot learning pipelines treat embodiment mismatch as a software problem. Collect data with one device, retarget the motion to another robot, and try to deal with the differences later. TwinDEX flips that idea around. This system is from a Chinese robotics company called X Square. Instead of starting with a generic data-collection device, the team co-designs a wearable three-finger end-effector and a robotic hand as two sides of the same learning system. They align the things that actually matter during manipulation: joint structure, link proportions, contact geometry, surface properties, visual appearance, sensing, and timing. That matters because dexterous manipulation is much more than matching hand poses. A few millimeters of contact error or a different force relationship can completely change how a task is executed. In the collection evaluation, TwinDEX delivered up to 5.3 times the effective throughput of on-robot teleoperation. It’s a useful reminder that scaling robot learning may not only be about bigger models and more data. Sometimes the hardware collecting the data is part of the data problem itself.

Alamin

48,112 Aufrufe • vor 1 Monat

I don't worry about the actual firmware stack. I would want every country to be using American chips so that the programmers there are held at bay by our technology. The Chinese are actually trying to develop this on their own. They are beating us in developing a larger electrical grid in China very quickly, because they don't have any permitting problems. And the other area that people have been talking about directly concerning these chips is they have much more compute power. They're building data centers at a rate that's four times ours. We have more data centers, but they're catching up very quickly.

I don't worry about the actual firmware stack. I would want every country to be using American chips so that the programmers there are held at bay by our technology. The Chinese are actually trying to develop this on their own. They are beating us in developing a larger electrical grid in China very quickly, because they don't have any permitting problems. And the other area that people have been talking about directly concerning these chips is they have much more compute power. They're building data centers at a rate that's four times ours. We have more data centers, but they're catching up very quickly.

Kevin O'Leary aka Mr. Wonderful

23,346 Aufrufe • vor 4 Monaten

Scale alone is not enough for AI data. Quality and complexity are equally critical. Excited to support all of these for LLM developers with Snorkel AI Data-as-a-Service, and to share our new leaderboard! — Our decade-plus of research and work in AI data has a simple point: scale alone is not enough. AI success is all about the quality, complexity, and distribution of data—in addition to volume. We’re excited to be powering leading LLM developers with Snorkel AI Expert Data-as-a-Service, our white glove service for custom, expert-level AI datasets—and to now preview some of what we’re building via our new Expert Data Leaderboard (🔗 in 🧵) + upcoming OSS dataset releases! Snorkel Expert Data-as-a-Service is built to meet the rapidly evolving data needs of the agentic AI world—where success is built on the quality, complexity, and distribution of datasets, in addition to size and scale. This kind of high-quality, frontier AI data can only come from a union of technology and human expertise. With Snorkel Expert Data-as-a-Service, we’re powering frontier LLM developers across agentic, expert knowledge, reasoning, coding, multi-modal, and other task types via the combination of these two key components: - (1) The Snorkel Expert Network: A global team of subject matter experts focused wholly on specialized knowledge–spanning thousands of topics in STEM/academic, vertical/professional, and consumer/lifestyle domains. - (2) Snorkel AI Data Development Platform: Our unique programmatic data curation and quality control platform, accelerating and improving expert authoring and review through principled techniques developed over the last decade of R&D. Now: we’re incredibly excited to showcase some of the power of Snorkel Expert Data-as-a-Service via the new Snorkel Leaderboard—putting frontier models to the test in complex, agentic, and reasoning settings inspired by real industry scenarios (not esoteric puzzles)! We’ll be releasing new leaderboards and accompanying expert-verified open source datasets (coming soon!) regularly. To start, we’re sharing three initial ones in preview: - SnorkelFinance: Q&A over financial documents requiring agentic tool-calling and reasoning - SnorkelUnderwrite: Agentic insurance tasks requiring industry-specific reasoning and tool use - SnorkelSequences: Mathematical tasks requiring compositional multi-step reasoning

Scale alone is not enough for AI data. Quality and complexity are equally critical. Excited to support all of these for LLM developers with Snorkel AI Data-as-a-Service, and to share our new leaderboard! — Our decade-plus of research and work in AI data has a simple point: scale alone is not enough. AI success is all about the quality, complexity, and distribution of data—in addition to volume. We’re excited to be powering leading LLM developers with Snorkel AI Expert Data-as-a-Service, our white glove service for custom, expert-level AI datasets—and to now preview some of what we’re building via our new Expert Data Leaderboard (🔗 in 🧵) + upcoming OSS dataset releases! Snorkel Expert Data-as-a-Service is built to meet the rapidly evolving data needs of the agentic AI world—where success is built on the quality, complexity, and distribution of datasets, in addition to size and scale. This kind of high-quality, frontier AI data can only come from a union of technology and human expertise. With Snorkel Expert Data-as-a-Service, we’re powering frontier LLM developers across agentic, expert knowledge, reasoning, coding, multi-modal, and other task types via the combination of these two key components: - (1) The Snorkel Expert Network: A global team of subject matter experts focused wholly on specialized knowledge–spanning thousands of topics in STEM/academic, vertical/professional, and consumer/lifestyle domains. - (2) Snorkel AI Data Development Platform: Our unique programmatic data curation and quality control platform, accelerating and improving expert authoring and review through principled techniques developed over the last decade of R&D. Now: we’re incredibly excited to showcase some of the power of Snorkel Expert Data-as-a-Service via the new Snorkel Leaderboard—putting frontier models to the test in complex, agentic, and reasoning settings inspired by real industry scenarios (not esoteric puzzles)! We’ll be releasing new leaderboards and accompanying expert-verified open source datasets (coming soon!) regularly. To start, we’re sharing three initial ones in preview: - SnorkelFinance: Q&A over financial documents requiring agentic tool-calling and reasoning - SnorkelUnderwrite: Agentic insurance tasks requiring industry-specific reasoning and tool use - SnorkelSequences: Mathematical tasks requiring compositional multi-step reasoning

Alex Ratner

495,851 Aufrufe • vor 1 Jahr

Training robot foundation models faces two key hurdles: how to get enough data to train an effective model, and how to make sure that new skills can be acquired quickly. The team at Rhoda AI believes that the answer is training Direct Video Action models from web data. Web data is plentiful, to the point where Rhoda can train their base model on hundreds of years of video data. And then, with the addition of robot data, they can quickly adapt it to new tasks with as little as 20 hours of in-domain data, performing complex, multi-step manipulation tasks with their purpose-built video foundation model. Tongzhou Mu 🤖🦾🦿 Eric Chan and Changan Chen joined us to talk more about their approach. Watch Episode #79 of RoboPapers, with Michael Cho - Rbt/Acc, Chris Paxton, and Jiafei Duan, to learn more!

Training robot foundation models faces two key hurdles: how to get enough data to train an effective model, and how to make sure that new skills can be acquired quickly. The team at Rhoda AI believes that the answer is training Direct Video Action models from web data. Web data is plentiful, to the point where Rhoda can train their base model on hundreds of years of video data. And then, with the addition of robot data, they can quickly adapt it to new tasks with as little as 20 hours of in-domain data, performing complex, multi-step manipulation tasks with their purpose-built video foundation model. Tongzhou Mu 🤖🦾🦿 Eric Chan and Changan Chen joined us to talk more about their approach. Watch Episode #79 of RoboPapers, with Michael Cho - Rbt/Acc, Chris Paxton, and Jiafei Duan, to learn more!

RoboPapers

24,728 Aufrufe • vor 5 Monaten

There is a new extremely viral theory going around that many of the data centers being built in America aren’t really data centers They’re actually modern day panopticon. A digital surveillance state Even Tucker Carlson thinks this is the case and brings it up with Kevin O'Leary There are some real reasons to question what we’re being told, one of the reasons is the size and scale Think about this China’s largest reported single data center campus is only about 200 football fields in size Compare that to the US where a single data center is being planned at 33,000 football fields. Not to mention there are tons of them that are massive being planned and built It really doesn’t make sense when you compare the population of China being bigger than the US This is a conspiracy, but it’s something to think about because we know America has been moving towards a surveillance state, that’s every real Keep paying attention and questioning things

There is a new extremely viral theory going around that many of the data centers being built in America aren’t really data centers They’re actually modern day panopticon. A digital surveillance state Even Tucker Carlson thinks this is the case and brings it up with Kevin O'Leary There are some real reasons to question what we’re being told, one of the reasons is the size and scale Think about this China’s largest reported single data center campus is only about 200 football fields in size Compare that to the US where a single data center is being planned at 33,000 football fields. Not to mention there are tons of them that are massive being planned and built It really doesn’t make sense when you compare the population of China being bigger than the US This is a conspiracy, but it’s something to think about because we know America has been moving towards a surveillance state, that’s every real Keep paying attention and questioning things

Wall Street Apes

178,613 Aufrufe • vor 4 Monaten

A model is only as good as its data, and we’ve long since exhausted the internet. From here on out, model progress is gated by data production. ’s Brendan (can/do) joined us at our Sovereign AI event to talk about how RL environments get built, and why your data might be your real moat: 00:00 Introduction 00:47 A short history of the data market: crowdsourcing to agentic data 02:29 What an RL environment is: worlds, apps, tasks 03:57 Why only humans can measure the frontier 05:35 Building verifiers is the hard part 06:44 Walkthrough: a real legal RL environment 08:18 Leaderboards — and what open weights change 09:45 Post-training results on Apex Agents 11:17 Three ways companies buy data 12:49 Q&A: How do you price data? 14:17 Q&A: What "data quality" actually means 16:42 Q&A: The misunderstanding about synthetic data 18:17 Q&A: Why RL environments now — and what comes after 21:20 Q&A: Can you scale rubric generation with models? 23:00 Q&A: RL environments for cyber defense 25:33 Q&A: Build data in-house or partner?

A model is only as good as its data, and we’ve long since exhausted the internet. From here on out, model progress is gated by data production. ’s Brendan (can/do) joined us at our Sovereign AI event to talk about how RL environments get built, and why your data might be your real moat: 00:00 Introduction 00:47 A short history of the data market: crowdsourcing to agentic data 02:29 What an RL environment is: worlds, apps, tasks 03:57 Why only humans can measure the frontier 05:35 Building verifiers is the hard part 06:44 Walkthrough: a real legal RL environment 08:18 Leaderboards — and what open weights change 09:45 Post-training results on Apex Agents 11:17 Three ways companies buy data 12:49 Q&A: How do you price data? 14:17 Q&A: What "data quality" actually means 16:42 Q&A: The misunderstanding about synthetic data 18:17 Q&A: Why RL environments now — and what comes after 21:20 Q&A: Can you scale rubric generation with models? 23:00 Q&A: RL environments for cyber defense 25:33 Q&A: Build data in-house or partner?

Pat Grady

66,691 Aufrufe • vor 1 Monat

We recently crossed 10,000 GitHub stars on Rerun 😍 That's great and all, but something bigger is happening in robotics and Physical AI right now: The open-source ecosystem has seen real acceleration over the last few years. More teams are sharing real infrastructure: datasets, models, simulators, tooling, and the unglamorous plumbing that makes iteration possible. Jan Florian Maas, Physical Intelligence, Hugging Face are just some of the names driving this forward. Visualization and data infra belong in that category. If your entire data loop depends on being able to query and inspect what your robots are doing, that layer can’t be fragile or vendor-controlled. It needs to be something you can introspect, build on, extend, and trust long-term. Huge thanks to everyone building open-source robotics. The next decade of Physical AI will be built in the open.

We recently crossed 10,000 GitHub stars on Rerun 😍 That's great and all, but something bigger is happening in robotics and Physical AI right now: The open-source ecosystem has seen real acceleration over the last few years. More teams are sharing real infrastructure: datasets, models, simulators, tooling, and the unglamorous plumbing that makes iteration possible. Jan Florian Maas, Physical Intelligence, Hugging Face are just some of the names driving this forward. Visualization and data infra belong in that category. If your entire data loop depends on being able to query and inspect what your robots are doing, that layer can’t be fragile or vendor-controlled. It needs to be something you can introspect, build on, extend, and trust long-term. Huge thanks to everyone building open-source robotics. The next decade of Physical AI will be built in the open.

Rerun

29,684 Aufrufe • vor 7 Monaten

Programmable Bandwidth is crypto’s next meta. Gm rent your spare Wi-Fi to AI. AI needs more data. AGI is coming. Are you ready? Bandwidth = how much data your connection moves per second. Residential IP = your home’s street address on the internet, trusted as human traffic that doesn’t get blocked. Together, Bandwidth + Residential IP = clean, high-trust traffic that data buyers and AI teams actually want. The community becomes the network. LLMs eat data. The more and the higher quality, the smarter they get. Owning the data pipeline = owning the power. Proof it’s valuable: big platforms license conversational & user-generated data for serious money. Many 8-figure+ deals are public, many more done under the table. Proprietary datasets are the real edge in this AI world. Why residential IPs matter: datacenter IPs get blocked by anti-scraping shields. Residential IP networks are resilient. Grass showed the playbook: idle bandwidth can pay twice 1️⃣ Web data collection via your node 2️⃣ Renting those nodes to GPU grids for training Data + compute = compounding revenue. Hub ( takes it further: aggregate community bandwidth → build a Residential IP Supernetwork → sell real-time data APIs to enterprises (from Meta to Web2 startups). The value loop: Community miners → Bandwidth pool → Enterprise feeds → Revenue → Rewards. If revenue outpaces incentives, everyone wins. Balancing it won’t be easy, time will tell. For miners: run on the network, earn points (likely $HUB at mainnet). Early participation can matter. Why DePIN? Community-run networks scale faster, are harder to censor, and more resilient than centralized systems. We are long programmable bandwidth thesis. In an agent-driven world, whoever controls the live data feed mints the money. DYOR. Register NOW

Programmable Bandwidth is crypto’s next meta. Gm rent your spare Wi-Fi to AI. AI needs more data. AGI is coming. Are you ready? Bandwidth = how much data your connection moves per second. Residential IP = your home’s street address on the internet, trusted as human traffic that doesn’t get blocked. Together, Bandwidth + Residential IP = clean, high-trust traffic that data buyers and AI teams actually want. The community becomes the network. LLMs eat data. The more and the higher quality, the smarter they get. Owning the data pipeline = owning the power. Proof it’s valuable: big platforms license conversational & user-generated data for serious money. Many 8-figure+ deals are public, many more done under the table. Proprietary datasets are the real edge in this AI world. Why residential IPs matter: datacenter IPs get blocked by anti-scraping shields. Residential IP networks are resilient. Grass showed the playbook: idle bandwidth can pay twice 1️⃣ Web data collection via your node 2️⃣ Renting those nodes to GPU grids for training Data + compute = compounding revenue. Hub ( takes it further: aggregate community bandwidth → build a Residential IP Supernetwork → sell real-time data APIs to enterprises (from Meta to Web2 startups). The value loop: Community miners → Bandwidth pool → Enterprise feeds → Revenue → Rewards. If revenue outpaces incentives, everyone wins. Balancing it won’t be easy, time will tell. For miners: run on the network, earn points (likely $HUB at mainnet). Early participation can matter. Why DePIN? Community-run networks scale faster, are harder to censor, and more resilient than centralized systems. We are long programmable bandwidth thesis. In an agent-driven world, whoever controls the live data feed mints the money. DYOR. Register NOW

Q42

39,768 Aufrufe • vor 1 Jahr

In the next 2-3 years, the skills you need to build an ETL process will fundamentally differ from anything we've done in the last decade. The EXTRACT and LOAD steps of an ETL pipeline have been solved forever. They are boring, and I suspect they will continue to be boring. But the TRANSFORM step is where the money is at (and where innovation will continue happening!) An ETL pipeline that can use Large Language Models effectively to transform data is a whole new ballgame. I recorded a video to talk about this. And more importantly, I talk about how you can build ETL pipelines to transform UNSTRUCTURED data (the kind of data that's really hard to handle).

In the next 2-3 years, the skills you need to build an ETL process will fundamentally differ from anything we've done in the last decade. The EXTRACT and LOAD steps of an ETL pipeline have been solved forever. They are boring, and I suspect they will continue to be boring. But the TRANSFORM step is where the money is at (and where innovation will continue happening!) An ETL pipeline that can use Large Language Models effectively to transform data is a whole new ballgame. I recorded a video to talk about this. And more importantly, I talk about how you can build ETL pipelines to transform UNSTRUCTURED data (the kind of data that's really hard to handle).

Santiago

56,412 Aufrufe • vor 1 Jahr

Chamath Warns: Your AI Data Isn’t as Private as You Think Today's statement from OpenAI on the Navier-Stokes Problem and whether it used recent user data to boost its own capabilities: “While unlikely, we cannot rule out that de-identified data derived from their usage of our products helped improve our models.” Here's what Chamath Palihapitiya had to say on AI's fragile privacy and the ZDR myth back in July: “Privacy in AI is very fragile and it's very brittle. There are all kinds of non-obvious data leak vectors lurking in AI. If you think that you're going to flip a ZDR switch, zero data retention, which is the magic term that the industry uses to tell you that everything's going to be okay. I think the answer and the message should be, ‘It's not going to be okay because you can't guarantee any of it.’ So the model companies, when they give you these zero data retention policies, are probably trying their best. But I think the reality is you are leaking information where you don't know it. And they, despite their best efforts, may still have trapdoors that they don't even know about until it's figured out by somebody else. You need an independent third-party layer to interface to these models to manage this exposure, because there are trapdoors everywhere.”

Chamath Warns: Your AI Data Isn’t as Private as You Think Today's statement from OpenAI on the Navier-Stokes Problem and whether it used recent user data to boost its own capabilities: “While unlikely, we cannot rule out that de-identified data derived from their usage of our products helped improve our models.” Here's what Chamath Palihapitiya had to say on AI's fragile privacy and the ZDR myth back in July: “Privacy in AI is very fragile and it's very brittle. There are all kinds of non-obvious data leak vectors lurking in AI. If you think that you're going to flip a ZDR switch, zero data retention, which is the magic term that the industry uses to tell you that everything's going to be okay. I think the answer and the message should be, ‘It's not going to be okay because you can't guarantee any of it.’ So the model companies, when they give you these zero data retention policies, are probably trying their best. But I think the reality is you are leaking information where you don't know it. And they, despite their best efforts, may still have trapdoors that they don't even know about until it's figured out by somebody else. You need an independent third-party layer to interface to these models to manage this exposure, because there are trapdoors everywhere.”

The All-In Podcast

69,508 Aufrufe • vor 26 Tagen

Yesterday I interviewed Sean Cai about AI data. This is essentially a guide for founders on how to sell data and RL envs to AI labs. "I've never seen a data contract get turned down by a top lab, if it's good quality data, for budget reasons." 00:00 What areas of data are underserved? 02:10 For bio data, is it real-world or purely digital? 04:21 For cyber data, which subsets are most underserved? 05:50 What is the sales process like? 07:04 Why would a lab not renew or increase their purchase volume? 10:13 When a researcher is exploring a new direction, what's the first step? 11:35 In robotics data, what do you view as underserved? 13:12 What does the initial data delivery look like, what format? 13:53 Do labs have more sophisticated internal setups for running environments? 14:32 Are the non-frontier labs buying off-the-shelf data from Anthropic / OpenAI vendors? 16:11 Do Anthropic data vendors put expiry timeframes on the exclusivity? 16:42 Are purchase decisions researcher-led? 17:41 Decagon, Sierra, Ramp: what kinds of data are they buying? 19:06 Long-term, when do labs still need to buy external data vs train on user traces? 21:15 Will end-vendor benchmarks shift to performance per dollar? 22:04 How many labs are spending at the 1B+/yr data level? 23:53 Delta between Anthropic's stated $1B and your 10-20B/lab number? 26:05 What makes inference providers / neoclouds a good fit to acquire RL env cos?

Yesterday I interviewed Sean Cai about AI data. This is essentially a guide for founders on how to sell data and RL envs to AI labs. "I've never seen a data contract get turned down by a top lab, if it's good quality data, for budget reasons." 00:00 What areas of data are underserved? 02:10 For bio data, is it real-world or purely digital? 04:21 For cyber data, which subsets are most underserved? 05:50 What is the sales process like? 07:04 Why would a lab not renew or increase their purchase volume? 10:13 When a researcher is exploring a new direction, what's the first step? 11:35 In robotics data, what do you view as underserved? 13:12 What does the initial data delivery look like, what format? 13:53 Do labs have more sophisticated internal setups for running environments? 14:32 Are the non-frontier labs buying off-the-shelf data from Anthropic / OpenAI vendors? 16:11 Do Anthropic data vendors put expiry timeframes on the exclusivity? 16:42 Are purchase decisions researcher-led? 17:41 Decagon, Sierra, Ramp: what kinds of data are they buying? 19:06 Long-term, when do labs still need to buy external data vs train on user traces? 21:15 Will end-vendor benchmarks shift to performance per dollar? 22:04 How many labs are spending at the 1B+/yr data level? 23:53 Delta between Anthropic's stated $1B and your 10-20B/lab number? 26:05 What makes inference providers / neoclouds a good fit to acquire RL env cos?

Chris Barber

132,580 Aufrufe • vor 4 Monaten

Better data for robot hands! 🤲🏼 Embodied reasoning is the bottleneck now. It’s the data we are missing. Specifically, understanding what the robot's hands are doing throughout a task. Perceptron AI new Egocentric offering tracks both hands through an entire video instead of guessing from sampled frames. That matters because a lot happens between frames. Hand positions shift. Contacts change. The robot adjusts grip. You miss all that if you're sampling. 0.280 semantic end-to-end F1 score on WGO-Bench vs 0.158 from the best pipelines built on Gemini Robotics-ER 1.6. Training a robot to manipulate objects requires understanding every moment of hand-object interaction. If your training data has gaps: missing frames, missed contacts, unclear hand positions, the policy learns from incomplete information. You end up with robots that work in demos but fail in real tasks. Continuous hand tracking gives you the full picture. The policy learns from cleaner, more complete data. That compounds into better real-world performance. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Better data for robot hands! 🤲🏼 Embodied reasoning is the bottleneck now. It’s the data we are missing. Specifically, understanding what the robot's hands are doing throughout a task. Perceptron AI new Egocentric offering tracks both hands through an entire video instead of guessing from sampled frames. That matters because a lot happens between frames. Hand positions shift. Contacts change. The robot adjusts grip. You miss all that if you're sampling. 0.280 semantic end-to-end F1 score on WGO-Bench vs 0.158 from the best pipelines built on Gemini Robotics-ER 1.6. Training a robot to manipulate objects requires understanding every moment of hand-object interaction. If your training data has gaps: missing frames, missed contacts, unclear hand positions, the policy learns from incomplete information. You end up with robots that work in demos but fail in real tasks. Continuous hand tracking gives you the full picture. The policy learns from cleaner, more complete data. That compounds into better real-world performance. ~~ ♻️ Join the weekly robotics newsletter, and never miss any news →

Lukas Ziegler

18,434 Aufrufe • vor 2 Monaten

🚨 GM GOT BUSTED SELLING DRIVERS’ LOCATION DATA — NOW WHAT THEY’RE REMOVING FROM NEW VEHICLES HAS PEOPLE FURIOUS GM got busted in a massive driver data scandal... and what it’s doing next has people asking what the hell the company is thinking. The FTC accused GM and OnStar of collecting and selling precise location and driving behavior data from millions of vehicles without adequate notice or consent. California then secured a $12.75 million settlement over GM’s sale of drivers’ location and driving data. Now GM has already ditched Apple CarPlay and Android Auto from many of its newer EVs, forcing drivers toward GM’s own software instead. After getting nailed over what happened with drivers’ data, GM wants MORE of your vehicle experience running through GM. And people are pissed. “So you sold my driving data and NOW you want even more control over my truck? LMFAO.” “Imagine paying $60K just to have the manufacturer decide which features you’re allowed to use 💀” “At this rate the vehicle won’t belong to you. You’ll just be paying GM rent to drive it.” GM says its software strategy is about creating a more integrated vehicle experience... not collecting more driver data. After GM got busted selling drivers’ location and driving data, why the hell would anyone trust them with even MORE control over their car?

🚨 GM GOT BUSTED SELLING DRIVERS’ LOCATION DATA — NOW WHAT THEY’RE REMOVING FROM NEW VEHICLES HAS PEOPLE FURIOUS GM got busted in a massive driver data scandal... and what it’s doing next has people asking what the hell the company is thinking. The FTC accused GM and OnStar of collecting and selling precise location and driving behavior data from millions of vehicles without adequate notice or consent. California then secured a $12.75 million settlement over GM’s sale of drivers’ location and driving data. Now GM has already ditched Apple CarPlay and Android Auto from many of its newer EVs, forcing drivers toward GM’s own software instead. After getting nailed over what happened with drivers’ data, GM wants MORE of your vehicle experience running through GM. And people are pissed. “So you sold my driving data and NOW you want even more control over my truck? LMFAO.” “Imagine paying $60K just to have the manufacturer decide which features you’re allowed to use 💀” “At this rate the vehicle won’t belong to you. You’ll just be paying GM rent to drive it.” GM says its software strategy is about creating a more integrated vehicle experience... not collecting more driver data. After GM got busted selling drivers’ location and driving data, why the hell would anyone trust them with even MORE control over their car?

HustleBitch

14,254 Aufrufe • vor 2 Tagen

James Fishback says Byron Donalds is LYING about data centers! Says this has nothing to do with China and it’s all about GREED. “There are 4,200 data centers in the United States. There are 470 in China. We've already won the AI war. My grandfather was a computer science major at Florida Atlantic University about 30 minutes south of here in Boca. And he would tell you that in the 1970s and 80s, computers were the size of Costco warehouses and now those same computers fit in our pocket. Why are we building an AI data center that in five years is going to be obsolete, is going to be one one hundredth of the size. And so I fear it's because a lot of the big land developers want to buy up cheap land, flip it for the next AI data center. So it's a greedy thing. It's a greed thing. And I don't want that to happen in our state because you know what? If you buy something and you screw up, you can return it. You could sell it. If you pave over citrus groves or the Everglades or a wetlands to build the data center, you cannot easily undo that. And so I don't want our farmers, I don't want our rural towns to be run roughshod over. My opponent, Byron Donalds, will say, well, what about the AI data center? Don't they have property rights? And I would say to him, what about the property rights of the people who live a hundred feet from the data center? Where there is EMF, where there is audible noise pollution, where there is the potential even for small radioactive waves? What happens in that case? Do they not have property rights? They don't have to live next to something like that.”

James Fishback says Byron Donalds is LYING about data centers! Says this has nothing to do with China and it’s all about GREED. “There are 4,200 data centers in the United States. There are 470 in China. We've already won the AI war. My grandfather was a computer science major at Florida Atlantic University about 30 minutes south of here in Boca. And he would tell you that in the 1970s and 80s, computers were the size of Costco warehouses and now those same computers fit in our pocket. Why are we building an AI data center that in five years is going to be obsolete, is going to be one one hundredth of the size. And so I fear it's because a lot of the big land developers want to buy up cheap land, flip it for the next AI data center. So it's a greedy thing. It's a greed thing. And I don't want that to happen in our state because you know what? If you buy something and you screw up, you can return it. You could sell it. If you pave over citrus groves or the Everglades or a wetlands to build the data center, you cannot easily undo that. And so I don't want our farmers, I don't want our rural towns to be run roughshod over. My opponent, Byron Donalds, will say, well, what about the AI data center? Don't they have property rights? And I would say to him, what about the property rights of the people who live a hundred feet from the data center? Where there is EMF, where there is audible noise pollution, where there is the potential even for small radioactive waves? What happens in that case? Do they not have property rights? They don't have to live next to something like that.”

Chris Nelson 🏝️🇺🇸

27,295 Aufrufe • vor 1 Monat

"One of the very confusing things about the models right now: how to reconcile the fact that they are doing so well on evals. And you look at the evals and you go, 'Those are pretty hard evals.' But the economic impact seems to be dramatically behind. There is [a possible] explanation. Back when people were doing pre-training, the question of what data to train on was answered, because that answer was everything. So you don't have to think if it's going to be this data or that data. When people do RL training, they say, 'Okay, we want to have this kind of RL training for this thing and that kind of RL training for that thing.' You say, 'Hey, I would love our model to do really well when we release it. I want the evals to look great. What would be RL training that could help on this task?' If you combine this with generalization of the models actually being inadequate, that has the potential to explain a lot of what we are seeing, this disconnect between eval performance and actual real-world performance"

"One of the very confusing things about the models right now: how to reconcile the fact that they are doing so well on evals. And you look at the evals and you go, 'Those are pretty hard evals.' But the economic impact seems to be dramatically behind. There is [a possible] explanation. Back when people were doing pre-training, the question of what data to train on was answered, because that answer was everything. So you don't have to think if it's going to be this data or that data. When people do RL training, they say, 'Okay, we want to have this kind of RL training for this thing and that kind of RL training for that thing.' You say, 'Hey, I would love our model to do really well when we release it. I want the evals to look great. What would be RL training that could help on this task?' If you combine this with generalization of the models actually being inadequate, that has the potential to explain a lot of what we are seeing, this disconnect between eval performance and actual real-world performance"

Dwarkesh Patel

502,761 Aufrufe • vor 10 Monaten