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。

We believe that evolutionary open-ended search and self-improving AI systems will be key to unlocking stronger AI capabilities. For a short overview, this video from Wes Roth discusses AlphaEvolve and our recent Darwin Gödel Machine. Full Video: Short Video:

Sakana AI

52,973 subscribers

23,821 次观看 • 1 年前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

6 条评论

Sakana AI 的头像

Sakana AI1 年前

The authors also recommend @MatthewBerman’s video for a nice introduction and walk through of the Darwin Gödel Machine paper. Video:

Coral AI News 的头像

Coral AI News1 年前

Coral AI is the most powerful AI for documents. See the difference yourself:

Ant A 的头像

Ant A1 年前

@WesRothMoney 👍

Rippa Sats 的头像

Rippa Sats1 年前

@WesRothMoney Will I be able to use this model? Better said, can I provide it certain Math to have it Rewrite itself?

Altiam Kabir 的头像

Altiam Kabir1 年前

@WesRothMoney Exciting potential for AI advancements!

Phil 的头像

Phil1 年前

@WesRothMoney If you don’t understand evey and all aspects of how it improved then you need to back off as you are playing with a fire that can impact humanity!!!

相关视频

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 次观看 • 1 年前

For our newest AI short film we used 100% open source models to create consistent and controllable creatures! We also used advanced video inpainting for two key shots. 👇Watch the short film below and see how this will change VFX! (Full tutorial and FREE workflows below)

For our newest AI short film we used 100% open source models to create consistent and controllable creatures! We also used advanced video inpainting for two key shots. 👇Watch the short film below and see how this will change VFX! (Full tutorial and FREE workflows below)

Mickmumpitz

46,292 次观看 • 11 个月前

“When AI Discovers the Next Transformer” Robert Lange (Sakana AI) joins Tim Scarfe (Machine Learning Street Talk) to discuss Shinka Evolve, a framework that combines LLMs with evolutionary algorithms to do open-ended program search. Full Video:

“When AI Discovers the Next Transformer” Robert Lange (Sakana AI) joins Tim Scarfe (Machine Learning Street Talk) to discuss Shinka Evolve, a framework that combines LLMs with evolutionary algorithms to do open-ended program search. Full Video:

Sakana AI

124,932 次观看 • 3 个月前

With our new UK data center now open, companies will be able to tap into our capacity powered by @NVIDIA Blackwell Ultra to develop new frontier models and AI solutions. Watch the short video from yesterday.

With our new UK data center now open, companies will be able to tap into our capacity powered by @NVIDIA Blackwell Ultra to develop new frontier models and AI solutions. Watch the short video from yesterday.

Nebius

53,468 次观看 • 8 个月前

The rant about why our current AI systems are useless and stupid from that video:

The rant about why our current AI systems are useless and stupid from that video:

Zane Little

64,463 次观看 • 4 个月前

A short video on how to use the free site for creating Ai video/Animation

A short video on how to use the free site for creating Ai video/Animation

Thelma Etuk

63,274 次观看 • 1 个月前

How to Create AI Films That Don't Look AI (Complete Tutorial) In this video, you'll learn how to to create a cinematic AI short film, without needing a big camera setup or crazy editing skills. You'll learn how to create simple story idea to full AI scenes using Seedance 2 and well planned cinematic shots to make the final video feel like a real short film. This information is for both new and existing Youtube creators, who want to elevate their video creation skills. Like, Retweet and Bookmark. Follow BIG CHRIS

How to Create AI Films That Don't Look AI (Complete Tutorial) In this video, you'll learn how to to create a cinematic AI short film, without needing a big camera setup or crazy editing skills. You'll learn how to create simple story idea to full AI scenes using Seedance 2 and well planned cinematic shots to make the final video feel like a real short film. This information is for both new and existing Youtube creators, who want to elevate their video creation skills. Like, Retweet and Bookmark. Follow BIG CHRIS

BIG CHRIS

74,677 次观看 • 21 天前

We do social media for a living, so we’re better equipped for short replies… that said we believe this video sums up our emotions 🥲

We do social media for a living, so we’re better equipped for short replies… that said we believe this video sums up our emotions 🥲

El Paso Chihuahuas

156,635 次观看 • 3 年前

Macro risk is rising, oil remains the key variable, and $NVDA earnings could decide this week’s AI trade. We turned today’s SoDEX Market Brief into a short video. Watch the key setup before the market moves. 👇 #SoDEX #MarketBrief #NVDA #AI #CryptoTrading

Macro risk is rising, oil remains the key variable, and $NVDA earnings could decide this week’s AI trade. We turned today’s SoDEX Market Brief into a short video. Watch the key setup before the market moves. 👇 #SoDEX #MarketBrief #NVDA #AI #CryptoTrading

SoSoValue

80,212 次观看 • 1 个月前

New No Priors with Pushmeet Kohli Matej Balog from DeepMind AlphaEvolve on: *ai co-scientists *improving matrix multiplication *reclaiming ~1% of google compute *can AI discover new knowledge? *evolutionary algorithms

New No Priors with Pushmeet Kohli Matej Balog from DeepMind AlphaEvolve on: ai co-scientists improving matrix multiplication reclaiming ~1% of google compute can AI discover new knowledge? *evolutionary algorithms

sarah guo

17,590 次观看 • 1 年前

Telecom infrastructure is becoming a distributed AI compute layer, unlocking smarter cities and more efficient operations. NVIDIA AI-RAN and NVIDIA Metropolis Blueprint for video search and summarization enable AI agents to run continuously at the edge. These agents analyze real-time data streams from connected sensors over 5G to power live monitoring and infrastructure inspection. ▶️ Watch the full video to learn more:

Telecom infrastructure is becoming a distributed AI compute layer, unlocking smarter cities and more efficient operations. NVIDIA AI-RAN and NVIDIA Metropolis Blueprint for video search and summarization enable AI agents to run continuously at the edge. These agents analyze real-time data streams from connected sensors over 5G to power live monitoring and infrastructure inspection. ▶️ Watch the full video to learn more:

NVIDIA AI

11,078 次观看 • 2 个月前

Today we are launching Dream Machine, our first AI model that generates cinematic and fluid videos from text instructions and images. I generated this 1-minute 60 fps video entirely from our model. Try Dream Machine → Join us →

Today we are launching Dream Machine, our first AI model that generates cinematic and fluid videos from text instructions and images. I generated this 1-minute 60 fps video entirely from our model. Try Dream Machine → Join us →

Dan Kondratyuk

38,992 次观看 • 2 年前

🎬 Midjourney + Kling + Udio: "In the Shadows" (AI movie short) I had a bit more time over the weekend and focused on improving consistency, sound effects, and transitions to make this video feel more cinematic. If I have more time, I hope to make more AI shorts like this! Images: Midjourney v7 Video: Kling AI Music: udio Editing: CapCut

🎬 Midjourney + Kling + Udio: "In the Shadows" (AI movie short) I had a bit more time over the weekend and focused on improving consistency, sound effects, and transitions to make this video feel more cinematic. If I have more time, I hope to make more AI shorts like this! Images: Midjourney v7 Video: Kling AI Music: udio Editing: CapCut

Everett World

33,608 次观看 • 1 年前

📽️ AI Agent for image-to-video video generation (beta) Combine with Pumpzy AI Agent - Add an image - Add a few words to generate video - Launch Pump Fun token right from AI chat Use-case: Make angry video & launch it on Pump Fun There is a short showcase from the AI chat 🔽

📽️ AI Agent for image-to-video video generation (beta) Combine with Pumpzy AI Agent - Add an image - Add a few words to generate video - Launch Pump Fun token right from AI chat Use-case: Make angry video & launch it on Pump Fun There is a short showcase from the AI chat 🔽

ZODs

14,673 次观看 • 1 年前

‘The Bakery Cat’ A short, wholesome AI-generated video for cat lovers

‘The Bakery Cat’ A short, wholesome AI-generated video for cat lovers

Lauren Sieckmann

61,272 次观看 • 1 年前

This is wild. Luma AI just dropped Dream Machine that generates AI video from text and image. Unlike Sora, it's open to public today. The quality is insane. 1. Kaku Drop 架空飴

This is wild. Luma AI just dropped Dream Machine that generates AI video from text and image. Unlike Sora, it's open to public today. The quality is insane. 1. Kaku Drop 架空飴

Min Choi

757,416 次观看 • 2 年前

Holy smokes! AI video just had its DeepSeek moment. LTX Video just got a massive update in both quality & speed. 13B parameters! It is a game-changing moment for AI video and... ...it’s open-source! See for yourself: 1. Fading Glory, a short I made in LTX Studio

Holy smokes! AI video just had its DeepSeek moment. LTX Video just got a massive update in both quality & speed. 13B parameters! It is a game-changing moment for AI video and... ...it’s open-source! See for yourself: 1. Fading Glory, a short I made in LTX Studio

Alex Patrascu

90,171 次观看 • 1 年前

We're changing our name. Captions is now Mirage. Today's most popular medium is short-form video. It's the fastest-growing format, it’s what people want, and it’s what brands need to grow. But unlike other forms of AI content like text or image, which can be generated and ready-to-use with a single prompt, generated video still requires a lot of manual work and editing to be ready to share. This is where the real opportunity lies. The way we see it, the real race for AI video hasn’t even begun. The world is waiting for a foundation model for video that can take an idea and produce a fully usable, cohesive output from scratch. So that’s what we’re building. We’re rebranding to Mirage to reflect our expanded vision for video, starting with the short-form format, through frontier AI research and models. Check it out at mirage dot app

We're changing our name. Captions is now Mirage. Today's most popular medium is short-form video. It's the fastest-growing format, it’s what people want, and it’s what brands need to grow. But unlike other forms of AI content like text or image, which can be generated and ready-to-use with a single prompt, generated video still requires a lot of manual work and editing to be ready to share. This is where the real opportunity lies. The way we see it, the real race for AI video hasn’t even begun. The world is waiting for a foundation model for video that can take an idea and produce a fully usable, cohesive output from scratch. So that’s what we’re building. We’re rebranding to Mirage to reflect our expanded vision for video, starting with the short-form format, through frontier AI research and models. Check it out at mirage dot app

Gaurav Misra

136,480 次观看 • 10 个月前