Video yükleniyor...

Video Yüklenemedi

Bu video yüklenirken bir sorun oluştu. Bu geçici bir ağ sorunundan kaynaklanıyor olabilir veya video kullanılamıyor olabilir.

Ana Sayfaya Dön

We’re finally building agents that learn just like this: playing directly in the real world to discover things on their own. It turns out the path to AGI involves a swarm of locally-hosted crustaceans. 🦞

hardmaru

404,065 subscribers

38,989 görüntüleme • 6 ay önce •via X (Twitter)

Eğitim Bilim & Teknoloji

Anya Rossi• Live Now

Private livecam show

0 Yorum

Yorum bulunmuyor

Orijinal gönderinin yorumları burada görünecek

Benzer Videolar

Demis Hassabis points out that today’s models don’t learn continuously, they are trained once and then deployed. Unlike humans, they don’t improve from experience or adapt after release. The next major leap, like “AlphaZero,” would involve self-learning systems that discover knowledge on their own. This ability to learn online and evolve in real time is a key missing component on the path to AGI. Solving this will be critical to building truly intelligent systems.

Demis Hassabis points out that today’s models don’t learn continuously, they are trained once and then deployed. Unlike humans, they don’t improve from experience or adapt after release. The next major leap, like “AlphaZero,” would involve self-learning systems that discover knowledge on their own. This ability to learn online and evolve in real time is a key missing component on the path to AGI. Solving this will be critical to building truly intelligent systems.

Wes Roth

39,238 görüntüleme • 8 ay önce

After operating in stealth for the last 18 months , we’re excited today to finally show the world what we’ve been working on. We believe we’re on a path to physical AGI with the launch of our brand new foundation model, the Direct Video Action (DVA) model.

Jagdeep Singh

265,618 görüntüleme • 5 ay önce

Everyone is obsessed with 🦞 OpenClaw🦞 and moltbook and the idea of autonomous agents running your life or building societies. But what happens when you scale that up? What if, instead of agents checking your calendar and posting on Reddit, you had hundreds of agents collaborating to construct the building blocks of life 🧬 from "first principles"? We created a Swarm of AIs to design never-before-seen-proteins, far outside of what Nature has produced. They negotiate, debate, and optimize locally to design proteins from scratch, with decentralized logic. No training required, pure emergence! These proteins 👇 were designed by a swarm of AIs!

Everyone is obsessed with 🦞 OpenClaw🦞 and moltbook and the idea of autonomous agents running your life or building societies. But what happens when you scale that up? What if, instead of agents checking your calendar and posting on Reddit, you had hundreds of agents collaborating to construct the building blocks of life 🧬 from "first principles"? We created a Swarm of AIs to design never-before-seen-proteins, far outside of what Nature has produced. They negotiate, debate, and optimize locally to design proteins from scratch, with decentralized logic. No training required, pure emergence! These proteins 👇 were designed by a swarm of AIs!

Markus J. Buehler

72,672 görüntüleme • 6 ay önce

New Paper! Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents A longstanding goal of AI research has been the creation of AI that can learn indefinitely. One path toward that goal is an AI that improves itself by rewriting its own code, including any code responsible for learning. That idea, known as a Gödel Machine, proposed by Jürgen Schmidhuber over two decades ago, is a hypothetical self-improving AI. It optimally solves problems by recursively rewriting its own code when it can mathematically prove a better strategy, making it a key concept in meta-learning or “learning to learn.” While the theoretical Gödel Machine promised provably beneficial self-modifications, its realization relied on an impractical assumption: that the AI could mathematically prove that a proposed change in its own code would yield a net improvement before adopting it. Sakana AI, in collaboration with Jeff Clune’s lab at UBC, proposes something more feasible: a system that harnesses the principles of open-ended algorithms like Darwinian evolution to search for improvements that empirically improve performance. We call the result the Darwin Gödel Machine. DGMs leverage foundation models to propose code improvements, and use recent innovations in open-ended algorithms to search for a growing library of diverse, high-quality AI agents. Applied to practical tasks, we implemented Darwin Gödel Machine as a self-improving coding agent that rewrites its own code to improve performance on programming tasks. It creates various self-improvements, such as a patch validation step, better file viewing, enhanced editing tools, generating and ranking multiple solutions to choose the best one, and adding a history of what has been tried before (and why it failed) when making new changes (see the attached video). We believe that Darwin Gödel Machines represent a concrete step towards AI systems that can autonomously gather their own stepping stones to learn and innovate forever!

New Paper! Darwin Godel Machine: Open-Ended Evolution of Self-Improving Agents A longstanding goal of AI research has been the creation of AI that can learn indefinitely. One path toward that goal is an AI that improves itself by rewriting its own code, including any code responsible for learning. That idea, known as a Gödel Machine, proposed by Jürgen Schmidhuber over two decades ago, is a hypothetical self-improving AI. It optimally solves problems by recursively rewriting its own code when it can mathematically prove a better strategy, making it a key concept in meta-learning or “learning to learn.” While the theoretical Gödel Machine promised provably beneficial self-modifications, its realization relied on an impractical assumption: that the AI could mathematically prove that a proposed change in its own code would yield a net improvement before adopting it. Sakana AI, in collaboration with Jeff Clune’s lab at UBC, proposes something more feasible: a system that harnesses the principles of open-ended algorithms like Darwinian evolution to search for improvements that empirically improve performance. We call the result the Darwin Gödel Machine. DGMs leverage foundation models to propose code improvements, and use recent innovations in open-ended algorithms to search for a growing library of diverse, high-quality AI agents. Applied to practical tasks, we implemented Darwin Gödel Machine as a self-improving coding agent that rewrites its own code to improve performance on programming tasks. It creates various self-improvements, such as a patch validation step, better file viewing, enhanced editing tools, generating and ranking multiple solutions to choose the best one, and adding a history of what has been tried before (and why it failed) when making new changes (see the attached video). We believe that Darwin Gödel Machines represent a concrete step towards AI systems that can autonomously gather their own stepping stones to learn and innovate forever!

hardmaru

104,854 görüntüleme • 1 yıl önce

The best thing to come out of what's promising to be the worst World Cup in history is that the world is finally going to discover that the United States is just the richest third world country on Earth.

The best thing to come out of what's promising to be the worst World Cup in history is that the world is finally going to discover that the United States is just the richest third world country on Earth.

Olga Bazova

19,627 görüntüleme • 2 ay önce

Introducing Theoriq’s AlphaSwarm! ⚡️ We’re thrilled to give developers access to a starter toolkit for building AI-powered trading agents in DeFi (AiFi)! 🦾 AlphaSwarm streamlines agent and swarm development, making it easier to build, execute, and refine collaborative agents that can analyze real-time market signals, and autonomously execute trades across multiple chains. Check out our latest blog for more details! 👇 You can explore the toolkit directly in Theoriq’s GitHub: Or if you’re a developer interested in building the future of AiFi, join our dev community in Discord:

Introducing Theoriq’s AlphaSwarm! ⚡️ We’re thrilled to give developers access to a starter toolkit for building AI-powered trading agents in DeFi (AiFi)! 🦾 AlphaSwarm streamlines agent and swarm development, making it easier to build, execute, and refine collaborative agents that can analyze real-time market signals, and autonomously execute trades across multiple chains. Check out our latest blog for more details! 👇 You can explore the toolkit directly in Theoriq’s GitHub: Or if you’re a developer interested in building the future of AiFi, join our dev community in Discord:

Theoriq

103,908 görüntüleme • 1 yıl önce

The world is severely underestimating the value of AI Gaming Agents in Web3 games Imagine you're playing The Sandbox Attack on Titan and you're playing with AI Agents that navigate like real humans Metaverse games are going to COOK!

The world is severely underestimating the value of AI Gaming Agents in Web3 games Imagine you're playing The Sandbox Attack on Titan and you're playing with AI Agents that navigate like real humans Metaverse games are going to COOK!

Brycent

22,217 görüntüleme • 1 yıl önce

hard to tell which ai agents are real right now and which ones are just hype we just officially went live on BNB Chain if bnb chain is the rails then we are the brain feeding them real world data instead of just noise so they can learn and trade on their own we just want agents that get smarter onchain

hard to tell which ai agents are real right now and which ones are just hype we just officially went live on BNB Chain if bnb chain is the rails then we are the brain feeding them real world data instead of just noise so they can learn and trade on their own we just want agents that get smarter onchain

NeoSoul

41,446 görüntüleme • 27 gün önce

Elon Musk on how AGI might actually emerge: “We’re not actually specifically focused on AGI. I’m simply saying that AGI seems likely to be an emergent property of what we’re doing” The reason is the massive real-world data Tesla is generating from autonomous cars and humanoid robots This is not internet data. This is real-world interaction driving on messy roads, navigating unpredictable environments, and interacting with people With tens or hundreds of millions of autonomous vehicles and potentially even more humanoids, the amount of real-world video and data being processed will be enormous From this, cars will become far better than human drivers, and humanoid robots will become increasingly indistinguishable from humans AGI might not come from a lab trying to build it directly. It may simply emerge from building useful robots at massive scale

Elon Musk on how AGI might actually emerge: “We’re not actually specifically focused on AGI. I’m simply saying that AGI seems likely to be an emergent property of what we’re doing” The reason is the massive real-world data Tesla is generating from autonomous cars and humanoid robots This is not internet data. This is real-world interaction driving on messy roads, navigating unpredictable environments, and interacting with people With tens or hundreds of millions of autonomous vehicles and potentially even more humanoids, the amount of real-world video and data being processed will be enormous From this, cars will become far better than human drivers, and humanoid robots will become increasingly indistinguishable from humans AGI might not come from a lab trying to build it directly. It may simply emerge from building useful robots at massive scale

X Freeze

21,256 görüntüleme • 2 ay önce

Here’s what it’s like to use @xAI’s Grok 4.20 (Beta) with 4 agents that just launched. Cool how all the agents think on their own and come up with the best answers together. Really like the new layout of things too.

Here’s what it’s like to use @xAI’s Grok 4.20 (Beta) with 4 agents that just launched. Cool how all the agents think on their own and come up with the best answers together. Really like the new layout of things too.

Sawyer Merritt

110,689 görüntüleme • 6 ay önce

How long does it take to learn how to drift like this in the real world?

How long does it take to learn how to drift like this in the real world?

ena

64,641 görüntüleme • 9 gün önce

Jamie Dimon, we’re coming for you. Kontigo is building the largest bank in the world. (A “full-stack” AI stablecoin bank). Most “stablecoin neobanks” are just Bridge wrappers with no real distribution. We’re building our own infrastructure to replace traditional banks. Our infra lets humans and agents transact on top of open networks (blockchains) and local stablecoins. “A startup building agents for law firms could just build the AI law firm.” We’re doing the same for banking. Disclaimer: Kontigo is not a bank. Banking services are provided by the freaking blockchain.

Jamie Dimon, we’re coming for you. Kontigo is building the largest bank in the world. (A “full-stack” AI stablecoin bank). Most “stablecoin neobanks” are just Bridge wrappers with no real distribution. We’re building our own infrastructure to replace traditional banks. Our infra lets humans and agents transact on top of open networks (blockchains) and local stablecoins. “A startup building agents for law firms could just build the AI law firm.” We’re doing the same for banking. Disclaimer: Kontigo is not a bank. Banking services are provided by the freaking blockchain.

Jesus A. Castillo F.

22,731 görüntüleme • 8 ay önce

What does JEPA actually learn? We can finally prove it 🌍 So excited to share our theory of identifiable World Models: LeJEPA recovers the latent variables of the world. Plan in the learned World Model as if it were real, same shortest path. 📄:

What does JEPA actually learn? We can finally prove it 🌍 So excited to share our theory of identifiable World Models: LeJEPA recovers the latent variables of the world. Plan in the learned World Model as if it were real, same shortest path. 📄:

David Klindt

217,428 görüntüleme • 2 ay önce

Reporter: Mr. President, are you planning to unveil a healthcare plan anytime soon in one place? Trump: Well, we’re looking at different alternatives. I mean, I like my plan the best: don’t give any money to the insurance companies—give it to the people directly. Let them go out and buy their own healthcare plan. And we’re looking at that. If that can work, we’re looking at it. That’s sort of taking it off. That’s what I like. Don’t give the money to the insurance companies. They go out and buy their own plan. You give the money to the people. I like it the best.

Reporter: Mr. President, are you planning to unveil a healthcare plan anytime soon in one place? Trump: Well, we’re looking at different alternatives. I mean, I like my plan the best: don’t give any money to the insurance companies—give it to the people directly. Let them go out and buy their own healthcare plan. And we’re looking at that. If that can work, we’re looking at it. That’s sort of taking it off. That’s what I like. Don’t give the money to the insurance companies. They go out and buy their own plan. You give the money to the people. I like it the best.

Acyn

621,147 görüntüleme • 8 ay önce

Today I am extremely proud to introduce the Fury Autonomous Vehicle Orchestrator 🇺🇸 For the past 12 months, the Scout AI team has been building this quietly. Now we’re finally ready to unveil it to the world Just like AI agents are transforming the digital world, Fury Orchestrator brings that same agentic intelligence into the physical world for the U.S. warfighter It is the first agentic system that can take a commander’s high-level mission intent, spoken or typed in natural language, and turn it into coordinated autonomous action across a mixed fleet of unmanned assets More in 🧵

Today I am extremely proud to introduce the Fury Autonomous Vehicle Orchestrator 🇺🇸 For the past 12 months, the Scout AI team has been building this quietly. Now we’re finally ready to unveil it to the world Just like AI agents are transforming the digital world, Fury Orchestrator brings that same agentic intelligence into the physical world for the U.S. warfighter It is the first agentic system that can take a commander’s high-level mission intent, spoken or typed in natural language, and turn it into coordinated autonomous action across a mixed fleet of unmanned assets More in 🧵

Colby Adcock

213,420 görüntüleme • 6 ay önce

built a small space on the internet not a portfolio, just things i’m building, thinking about, and collecting like a journal site → adayinalife[dot]vercel[dot]app (check thread) how it works: i dump thoughts into poke ai Poke → it structures them in Notion → site pulls and turns it into a journal it’s kind of like my own twitter, just more personal hover to peek, click to go deeper built with vite + Cursor, deployed on Vercel this will always be a work in progress i’ll keep adding things as i go ( open to ideas and building something together and might make it public to make their own journals )

built a small space on the internet not a portfolio, just things i’m building, thinking about, and collecting like a journal site → adayinalife[dot]vercel[dot]app (check thread) how it works: i dump thoughts into poke ai Poke → it structures them in Notion → site pulls and turns it into a journal it’s kind of like my own twitter, just more personal hover to peek, click to go deeper built with vite + Cursor, deployed on Vercel this will always be a work in progress i’ll keep adding things as i go ( open to ideas and building something together and might make it public to make their own journals )

anurag 🌻

14,466 görüntüleme • 4 ay önce

Infrastructure defines what scales. Electricity enabled industry. The internet enabled software. Now AI is moving toward systems that run continuously, across data, execution, and feedback. 💡At OptimAI, we’re building that infrastructure. A decentralized search layer for real-time data. A network of nodes powering continuous retrieval and validation. Agents that execute tasks directly on your device. This is not theoretical - we’re already seeing it in production, with millions of nodes and growing usage across the network. We’re putting together a short documentary to show how this shift is happening, and where it leads next. If you’re building, exploring, or just getting started - this is the layer to pay attention to. Welcome to OptimAI.

Infrastructure defines what scales. Electricity enabled industry. The internet enabled software. Now AI is moving toward systems that run continuously, across data, execution, and feedback. 💡At OptimAI, we’re building that infrastructure. A decentralized search layer for real-time data. A network of nodes powering continuous retrieval and validation. Agents that execute tasks directly on your device. This is not theoretical - we’re already seeing it in production, with millions of nodes and growing usage across the network. We’re putting together a short documentary to show how this shift is happening, and where it leads next. If you’re building, exploring, or just getting started - this is the layer to pay attention to. Welcome to OptimAI.

OptimAI Network

17,472 görüntüleme • 4 ay önce

Hi #GodotEngine folks! 😀 I finally updated the README of my starter kit to show all the things to discover in it! 🚀Link in the bio 🚀 (some videos aren't playing on the recording below, my bad! )

Hi #GodotEngine folks! 😀 I finally updated the README of my starter kit to show all the things to discover in it! 🚀Link in the bio 🚀 (some videos aren't playing on the recording below, my bad! )

Nodragem

61,536 görüntüleme • 1 yıl önce